{ "updated": "2026-10-09", "models": [ { "checked": "2026-10-09", "discovered": "2026-10-09", "maxOutput": null, "tools": null, "slug": "blama-8b", "providerId": "allenai/Llama-3-Blama-8B", "identityLabel": "checkpoint", "name": "Blama 8B", "provider": "Ai2", "announced": "2026-10-07", "released": null, "verification": "Byte-level research checkpoint", "context": null, "modalities": [ "text" ], "sourceUrl": "https://huggingface.co/allenai/Llama-3-Blama-8B", "officialUrl": "https://allenai.org/blog/bolmo-nature", "note": "A research model that reads and writes bytes instead of ordinary subword tokens. Ai2\u2019s October 7 announcement highlights this new checkpoint; repository creation and modification dates are in August, so announcement and artifact dates stay separate.", "capabilityNote": "This is a base autoregressive text checkpoint, rather than a documented tool-using chat assistant. Its card refers to custom Transformers code and counts max_new_tokens in bytes.", "contextNote": "The configuration\u2019s max_position_embeddings is 8,192. That internal position allowance is not treated as a standard token-context limit. Byte budgets are not directly comparable to the token budgets of hosted chat models.", "accessNote": "The card declares Llama 3 Community License. Blama\u2019s card also uses the older Llama-B name and links; the current repository ID is kept exact. Stage 1 artifacts are training starting points, not extra completed base-model releases. Parent Qwen/Llama scores are not transferred.", "pricingNote": "Hosted prices, output caps and reasoning controls for this exact checkpoint are unverified. Downloadable weights do not make inference free.", "catalogId": null, "catalogAdded": null, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "rates": { "input": null, "output": null, "cached": null, "write": null, "overrides": [] }, "independentEvidence": [] }, { "checked": "2026-10-09", "discovered": "2026-10-09", "maxOutput": null, "tools": null, "slug": "bwen-8b", "providerId": "allenai/Bwen-8B", "identityLabel": "checkpoint", "name": "Bwen 8B", "provider": "Ai2", "announced": "2026-10-07", "released": null, "verification": "Byte-level research checkpoint", "context": null, "modalities": [ "text" ], "sourceUrl": "https://huggingface.co/allenai/Bwen-8B", "officialUrl": "https://allenai.org/blog/bolmo-nature", "note": "A research model that reads and writes bytes instead of ordinary subword tokens. Ai2\u2019s October 7 announcement highlights this new checkpoint; repository creation and modification dates are in August, so announcement and artifact dates stay separate.", "capabilityNote": "This is a base autoregressive text checkpoint, rather than a documented tool-using chat assistant. Its card refers to custom Transformers code and counts max_new_tokens in bytes.", "contextNote": "The configuration\u2019s max_position_embeddings is 32,768. That internal position allowance is not treated as a standard token-context limit. Byte budgets are not directly comparable to the token budgets of hosted chat models.", "accessNote": "The card declares Apache 2.0. Blama\u2019s card also uses the older Llama-B name and links; the current repository ID is kept exact. Stage 1 artifacts are training starting points, not extra completed base-model releases. Parent Qwen/Llama scores are not transferred.", "pricingNote": "Hosted prices, output caps and reasoning controls for this exact checkpoint are unverified. Downloadable weights do not make inference free.", "catalogId": null, "catalogAdded": null, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "rates": { "input": null, "output": null, "cached": null, "write": null, "overrides": [] }, "independentEvidence": [] }, { "checked": "2026-10-09", "discovered": "2026-10-09", "maxOutput": null, "tools": false, "slug": "liquid-d1-omni-600m", "providerId": "LiquidAI/d1-omni-600M", "identityLabel": "checkpoint", "name": "Liquid d1 Omni 600M", "provider": "Liquid AI", "released": "2026-10-07", "verification": "Open-weight decision checkpoint", "specialized": "decision", "deployment": "checkpoint", "context": 16384, "questionLimit": null, "modalities": [ "text", "image", "audio" ], "sourceUrl": "https://huggingface.co/LiquidAI/d1-omni-600M", "officialUrl": "https://www.liquid.ai/blog/d1-open", "outputType": "Choice, Score and Noul probabilities without generated output tokens.", "note": "An experimental local decision checkpoint for text with images or audio. It can route spoken requests without generating a transcript or free-form answer.", "capabilityNote": "Use the checkpoint\u2019s supplied decision interface and matching preprocessing. Generic Hugging Face chat examples do not establish a supported generative chat contract.", "contextNote": "The request limit is 16,384 combined text/image/audio positions. With images, state and question text are cut to 896 tokens. Audio clips are cut at 30 seconds, and training covers English speaker\u2013assistant requests.", "accessNote": "Weights use the LFM Open License v1.0, with a commercial-use revenue limitation; review the exact checkpoint license before deployment. Provider benchmark and device timings are not independent measurements. Hosted token prices are unknown; local inference has hardware and operating costs.", "pricingNote": "No public hosted per-token quote is verified for this exact checkpoint. It is not the hosted d1 API and does not inherit its $0.04 rate.", "catalogId": null, "catalogAdded": null, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "rates": { "input": null, "output": null, "cached": null, "write": null, "overrides": [] }, "independentEvidence": [] }, { "checked": "2026-10-09", "discovered": "2026-10-09", "maxOutput": null, "tools": false, "slug": "liquid-d1-3b", "providerId": "LiquidAI/d1-3B", "identityLabel": "checkpoint", "name": "Liquid d1 3B", "provider": "Liquid AI", "released": "2026-10-07", "verification": "Open-weight decision checkpoint", "specialized": "decision", "deployment": "checkpoint", "context": 32768, "questionLimit": null, "modalities": [ "text", "image" ], "sourceUrl": "https://huggingface.co/LiquidAI/d1-3B", "officialUrl": "https://www.liquid.ai/blog/d1-open", "outputType": "Choice, Score and Noul probabilities without generated output tokens.", "note": "A local decision checkpoint for text and images. It returns fixed answers in one pass rather than writing a response.", "capabilityNote": "Use the checkpoint\u2019s supplied decision interface and matching preprocessing. Generic Hugging Face chat examples do not establish a supported generative chat contract.", "contextNote": "The model card documents 32,768 context tokens for d1-3B.", "accessNote": "Weights use the LFM Open License v1.0, with a commercial-use revenue limitation; review the exact checkpoint license before deployment. Provider benchmark and device timings are not independent measurements. Hosted token prices are unknown; local inference has hardware and operating costs.", "pricingNote": "No public hosted per-token quote is verified for this exact checkpoint. It is not the hosted d1 API and does not inherit its $0.04 rate.", "catalogId": null, "catalogAdded": null, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "rates": { "input": null, "output": null, "cached": null, "write": null, "overrides": [] }, "independentEvidence": [] }, { "checked": "2026-10-09", "discovered": "2026-10-09", "maxOutput": null, "tools": false, "slug": "liquid-d1", "providerId": "d1", "name": "Liquid d1", "provider": "Liquid AI", "released": "2026-10-05", "verification": "Hosted decision model", "specialized": "decision", "context": null, "questionLimit": null, "modalities": [ "text", "image" ], "sourceUrl": "https://www.liquid.ai/blog/d1-decision-model", "officialUrl": "https://docs.liquid.ai/lfm/models/decision-models", "outputType": "Choice distributions, ordered Score ratings and yes/no Noul probabilities; no free-form text.", "documentedRates": { "input": 0.04, "output": 0, "cached": null, "label": "Direct decision API base rate \u00b7 USD" }, "note": "Use this hosted model when your task has a fixed set of answers, such as routing a ticket or checking a record. It accepts images directly; it is distinct from the two downloadable d1 checkpoints.", "capabilityNote": "The native /decisions/v1/systemone endpoint accepts text and images under model d1. Context, maximum questions and calibration on your own labels remain unverified.", "contextNote": "Gateway lists liquid/d1 with 65,536 context and vision metadata. The provider announcement says routed access is text-only for now, so image support must be checked for the exact route. max_tokens=0 describes typed answers, not zero capacity.", "pricingNote": "The published base input rate is $0.04 per million. Each question is billed as its own prompt, including its text and all images. Images count as 1.5 tokens per 32\u00d732-pixel patch. Use the sum of billed tokens across questions in the calculator.", "billingNote": "Each question repeats its billable prompt and images. Zero generated tokens means no output-token charge, not free requests.", "accessNote": "Provider comparisons use one run per model on six applications, default chat-model effort, list prices and no cache discounts. They are vendor evidence, not a matched independent quality or latency ranking.", "catalogId": null, "catalogAdded": null, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "rates": { "input": null, "output": null, "cached": null, "write": null, "overrides": [] }, "independentEvidence": [] }, { "checked": "2026-10-09", "discovered": "2026-10-09", "maxOutput": null, "tools": null, "slug": "callstack-apex", "providerId": "apex-20261001", "identityLabel": "pinned native model", "name": "Callstack Apex", "provider": "Callstack", "released": "2026-10-01", "verification": "Specialized hosted model", "context": 262144, "modalities": [ "text" ], "sourceUrl": "https://apex.callstack.com/models", "officialUrl": "https://www.callstack.com/blog/apex-agetic-coding-for-mobile-and-web-spcialied-in-react", "capabilitySourceUrl": "https://ai-gateway.vercel.sh/v1/models", "documentedRates": { "input": 0.5, "output": 3, "cached": 0.2, "label": "AI Gateway offer \u00b7 USD" }, "note": "A model trained for React Native and React engineering, with Next.js use described by the provider. The native pinned version is October 1; Gateway availability is a separate October 8 offer. Comparable independent coding results are pending.", "capabilityNote": "The native contract lists text input and a 262,144-token context. Use apex-20261001 for a stable snapshot; apex is a moving alias. Gateway separately lists callstack/apex with text/image input, tools and 131,072 maximum output; those are hosted-route declarations, not verified native capabilities.", "gatewayEffortOptions": [ "none", "low", "medium", "xhigh" ], "contextNote": "The native maximum output and effort default remain unverified. Native text-only documentation differs from Gateway\u2019s image-input declaration; test the route you will use before relying on screenshots.", "pricingNote": "The Gateway catalog quotes $0.50 input, $0.20 cache reads and $3 output per million, checked October 9. The linked native pricing page returns 404; these are not verified direct-provider list prices.", "accessNote": "The provider identifies a licensed Qwen base without publishing an exact base checkpoint. Weights are not publicly distributed. Hosted and commercial self-hosting access are separate choices; no Qwen benchmark is transferred.", "catalogId": null, "catalogAdded": null, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "rates": { "input": null, "output": null, "cached": null, "write": null, "overrides": [] }, "independentEvidence": [] }, { "slug": "gpt-6-luna-decisions", "providerId": "gpt-6-luna", "identityLabel": "Decisions API model", "name": "GPT-6 Luna Decisions", "provider": "OpenAI", "released": "2026-10-06", "verification": "Public beta API offer", "specialized": "decision", "context": null, "maxOutput": null, "questionLimit": null, "modalities": [ "text", "image" ], "tools": false, "sourceUrl": "https://developers.openai.com/api/docs/guides/decisions", "officialUrl": "https://developers.openai.com/api/docs/changelog", "documentedRates": { "input": 0.1, "output": 0, "cached": 0, "write": 0, "label": "Direct Decisions API \u00b7 USD" }, "note": "A dedicated typed-answer API using GPT-6 Luna. It returns predicates, choices and rubric scores with probabilities; chat benchmark results do not establish decision accuracy.", "outputType": "Typed predicates, choices and rubric scores with probabilities and confidence fields.", "capabilityNote": "Use POST /v1/decisions with model gpt-6-luna. Input supports text and images. Native context and question-count caps and an effort contract are not verified in the reviewed guide.", "contextNote": "Gateway lists 1,050,000 context tokens for its route. Its zero max_tokens is typed-output metadata, not proof of zero capacity or a native API limit.", "pricingNote": "The base input rate is $0.10 per million tokens, with no separate cache-read, cache-write or output-token charge. Regional processing and long-context input premiums apply. Other GPT-6 Luna APIs use their own rates.", "billingNote": "Base input charge only; regional and long-context premiums can increase it. Cache reads/writes and output tokens have no separate charge.", "accessNote": "This is a public beta API offer, rather than a new base model. The publisher\u2019s speed comparison is not a matched independent latency result.", "discovered": "2026-10-07", "checked": "2026-10-09", "catalogId": null, "catalogAdded": null, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "rates": { "input": null, "output": null, "cached": null, "write": null, "overrides": [] }, "independentEvidence": [] }, { "slug": "astabrief-8b", "providerId": "allenai/AstaBrief_8B", "identityLabel": "checkpoint", "name": "AstaBrief 8B", "provider": "Ai2", "released": "2026-10-02", "verification": "Open-weight research model", "specialized": "research", "context": 40960, "maxOutput": null, "modalities": [ "text" ], "tools": null, "sourceUrl": "https://huggingface.co/allenai/AstaBrief_8B", "officialUrl": "https://allenai.org/blog/astabrief", "note": "A model for turning a research question and supplied literature excerpts into a cited report. Its published development studies do not compare today\u2019s frontier models.", "capabilityNote": "Use the recommended report prompt and provide retrieved literature. This checkpoint is a report writer; its upstream retrieval pipeline is separate. The configuration sets max_position_embeddings to 40,960, not a verified hosted service limit.", "contextNote": "40,960 is checkpoint configuration evidence. The example\u2019s 4,096 max_tokens is a request setting, not a published maximum output allowance.", "accessNote": "The checkpoint declares Apache 2.0 and can be run locally. Hosted per-token prices are not verified. Vendor pipeline timings and 2025 development comparisons are not imported into general quality or speed rankings.", "discovered": "2026-10-07", "checked": "2026-10-09", "catalogId": null, "catalogAdded": null, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "rates": { "input": null, "output": null, "cached": null, "write": null, "overrides": [] }, "independentEvidence": [] }, { "slug": "kolibri-1", "providerId": "Aleph-Alpha/Kolibri-1", "identityLabel": "checkpoint", "name": "Kolibri 1", "provider": "Aleph Alpha", "released": "2026-10-03", "verification": "Open-weight checkpoint", "context": 1048576, "maxOutput": null, "modalities": [ "text" ], "tools": true, "supportedEfforts": [ "none", "low", "medium", "high" ], "sourceUrl": "https://huggingface.co/Aleph-Alpha/Kolibri-1", "officialUrl": "https://aleph-alpha.com/en/news/kolibri-sovereign-ai-made-in-germany/", "note": "An open-weight German and English reasoning model with tool support. Downloadable weights are available; matched independent general-task scores and hosted token prices are pending.", "contextNote": "The card states 262,144 native context tokens and validates extension to 1,048,576. It recommends at most 262,144 for complex tasks and serving efficiency. These are checkpoint claims, not a hosted API guarantee.", "capabilityNote": "This exact FP8 checkpoint supports text and tool calls through its serving parser. Thinking levels are configured through the chat template; an omitted default is unverified.", "accessNote": "The checkpoint declares Apache 2.0. Running it yourself requires hardware, serving and operational work; it does not establish a zero-cost hosted API.", "discovered": "2026-10-07", "checked": "2026-10-09", "catalogId": null, "catalogAdded": null, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "rates": { "input": null, "output": null, "cached": null, "write": null, "overrides": [] }, "independentEvidence": [] }, { "slug": "reka-rho-1", "providerId": null, "name": "Reka Rho-1", "provider": "Reka", "announced": "2026-10-05", "released": null, "verification": "Research preview", "context": null, "maxOutput": null, "modalities": [ "text", "image", "video" ], "tools": null, "sourceUrl": "https://reka.ai/news/rho-1-collapsing-the-multimodal-stack", "officialUrl": "https://reka.ai/labs", "note": "A research preview that combines text, image and video understanding and generation. General task quality, latency and billing are not independently verified here.", "capabilityNote": "The provider demonstrates one multimodal model and a shared state. The demonstration is not an API tool-calling contract. Public API ID, context and output limits, effort controls and prices remain unverified.", "accessNote": "The provider offers research collaboration. A public announcement does not establish ordinary API-account access.", "discovered": "2026-10-07", "checked": "2026-10-09", "catalogId": null, "catalogAdded": null, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "rates": { "input": null, "output": null, "cached": null, "write": null, "overrides": [] }, "independentEvidence": [] }, { "slug": "reflection-beam", "providerId": "Beam-501B-A23B", "name": "Reflection Beam", "provider": "Reflection", "released": "2026-10-05", "verification": "Early access beta", "context": 262144, "maxOutput": 131072, "modalities": [ "text" ], "tools": true, "supportedEfforts": [ "low", "medium", "high", "xhigh", "max" ], "defaultEffort": "medium", "sourceUrl": "https://developers.reflection.ai/models", "officialUrl": "https://reflection.ai/blog/introducing-beam", "note": "A text model for coding and tool-based work, currently behind beta access. Its API limits and five effort settings are documented; comparable independent results and public token prices are not verified.", "accessNote": "New accounts join a waitlist and can create keys after access is enabled. Weights and developer artifacts are promised later in October; they are not available simply because the model is announced.", "capabilityNote": "The native API documents 262,144 combined context tokens and up to 131,072 output tokens. Reasoning is always on; preserve reasoning_content when replaying tool calls.", "pricingNote": "Public per-token prices are not verified. Check your account contract; neither beta access nor planned open weights establishes free inference.", "discovered": "2026-10-07", "checked": "2026-10-09", "catalogId": null, "catalogAdded": null, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "rates": { "input": null, "output": null, "cached": null, "write": null, "overrides": [] }, "independentEvidence": [] }, { "slug": "gpt-rosalind-research", "providerId": "gpt-rosalind-research", "name": "GPT-Rosalind Research", "provider": "OpenAI", "discovered": "2026-10-05", "released": null, "verification": "Restricted research access", "specialized": "research", "context": null, "maxOutput": null, "modalities": [], "tools": null, "sourceUrl": "https://developers.openai.com/api/docs/pricing#specialized-models", "officialUrl": "https://developers.openai.com/api/docs/models", "note": "A specialized life-sciences reasoning model for approved research organizations. OpenAI lists October 5 as its billing start; this is a pricing milestone, not a new model launch. Comparable independent benchmarks are pending.", "documentedRates": { "input": 5, "output": 25, "cached": 0.5, "label": "Direct restricted API \u00b7 USD" }, "accessNote": "OpenAI\u2019s September 8 access notice requires approval through its trusted-access program and limits use to approved internal research. A standard API account does not establish access.", "capabilityNote": "OpenAI describes this as a life-sciences reasoning model. The reviewed sources do not establish its context window, output limit, input formats, tools or effort settings.", "pricingNote": "Published standard rates per million tokens: $5 input, $0.50 cached input and $25 output. Billing begins October 5, 2026; an exact cutoff time and timezone are not published. OpenAI says cache-write pricing does not apply. Task costs and measured speed are unavailable.", "dateNote": "Approved research access was announced September 8, 2026. The original model launch date is unverified.", "checked": "2026-10-09", "catalogId": null, "catalogAdded": null, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "rates": { "input": null, "output": null, "cached": null, "write": null, "overrides": [] }, "independentEvidence": [] }, { "slug": "jev", "providerId": "typesafe-ai/jev", "name": "Jev", "provider": "TypeSafe AI", "verification": "Hosted decision route", "identityLabel": "AI Gateway route", "discovered": "2026-10-04", "released": null, "announced": "2026-09-16", "context": 32000, "maxOutput": null, "modalities": [ "text" ], "tools": false, "specialized": "decision", "questionLimit": null, "outputType": "Typed choices, rubric scores and Boolean probabilities", "sourceUrl": "https://vercel.com/ai-gateway/models/jev", "officialUrl": "https://docs.typesafe.ai/introduction", "capabilitySourceUrl": "https://ai-gateway.vercel.sh/v1/models", "note": "Newly covered TypeSafe System One decision endpoint. The Gateway announcement is dated September 16; its catalog release field is September 15, so a base-model launch date is not inferred. Independent matched decision quality and calibration are pending.", "contextNote": "Gateway declares 32,000 tokens for shared state plus the longest question, and 64,000 total request tokens. This is not a 64K single-state context. The maximum question count is unverified.", "capabilityNote": "The exact evaluation contract accepts text state and typed Choice, Score and Boolean questions. It does not stream free-form output; no independent speed advantage or calibrated confidence is inferred.", "pricingNote": "The Gateway API quotes $0.042 per million billed input tokens and zero output-token price, observed October 4. Its model page rounds the input quote to $0.04; the calculator preserves the exact catalog quote. Caching and other fees remain unverified.", "documentedRates": { "input": 0.042, "output": 0, "cached": null, "label": "AI Gateway decision offer \u00b7 USD" }, "billingNote": "Gateway explicitly quotes zero output-token price; cache and other fees are not verified.", "accessNote": "Use typesafe-ai/jev through the TypeSafe evaluation endpoint, not as a free-form chat model. The contract has no free-form output-token maximum; catalog max_tokens=0 is not a zero-capacity limit.", "checked": "2026-10-09", "catalogId": null, "catalogAdded": null, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "rates": { "input": null, "output": null, "cached": null, "write": null, "overrides": [] }, "independentEvidence": [] }, { "slug": "deepseek-v4-1-flash-fast", "providerId": "deepseek/deepseek-v4.1-flash-fast", "name": "DeepSeek V4.1 Flash Fast", "provider": "DeepSeek", "verification": "Hosted serving variant", "identityLabel": "AI Gateway route", "discovered": "2026-10-03", "released": null, "context": 1048576, "maxOutput": 1000000, "modalities": [ "text", "image" ], "tools": true, "sourceUrl": "https://vercel.com/ai-gateway/models/deepseek-v4.1-flash-fast", "capabilitySourceUrl": "https://ai-gateway.vercel.sh/v1/models", "gatewayEffortOptions": [ "none", "low", "medium", "high" ], "note": "AI Gateway lists a speed-optimized V4.1 Flash deployment. This is a serving variant; no base-model scores or vendor speed claims become measured results here.", "capabilityNote": "The exact Gateway API declares text/image input, tool use, reasoning controls, a 1,048,576-token context and 1,000,000 output tokens. Native mappings and independent variant measurements are pending.", "contextNote": "Limits belong to this hosted route and may differ from another deployment of V4.1 Flash.", "pricingNote": "Gateway rates observed October 4 are $0.30 input, $1.20 output and $0.006 cache read per million tokens. These are routed quotes, separate from direct-provider price schedules. This is a historical offer, not a currently verified quote.", "documentedRates": { "input": 0.3, "output": 1.2, "cached": 0.006, "label": "AI Gateway routed offer \u00b7 USD" }, "accessNote": " The route is absent from the October 9 Gateway catalog and its model page returns 404; verify access before planning new work. Historical price and capability observations below retain their original scope.", "checked": "2026-10-09", "catalogId": null, "catalogAdded": null, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "rates": { "input": null, "output": null, "cached": null, "write": null, "overrides": [] }, "independentEvidence": [] }, { "slug": "laya", "providerId": "convaiinnovations/laya", "name": "Laya", "provider": "Convai Innovations", "verification": "Hosted decision route", "identityLabel": "AI Gateway route", "discovered": "2026-10-03", "released": null, "context": 8192, "maxOutput": null, "modalities": [ "text" ], "tools": false, "specialized": "decision", "questionLimit": null, "outputType": "Yes/no probabilities, bounded choices and rubric scores", "sourceUrl": "https://vercel.com/ai-gateway/models/laya", "officialUrl": "https://vercel.com/changelog/laya-decision-model-now-available-on-ai-gateway-free-through-october-31", "note": "Typed decision model hosted on AI Gateway. Access was announced October 1; the base checkpoint launch date and independent matched quality are unverified.", "contextNote": "The exact Gateway API lists 8,192 tokens. A maximum question count is unverified; this is not a guarantee that every state plus question fits.", "capabilityNote": "The hosted evaluation API documents text state and typed yes/no, choice and score questions. No hosted image support or independent calibration sweep is established.", "pricingNote": "Boundless serves both standard and -free routes free through October 31, 2026. Select Boundless for this promotion. Afterward the standard route bills and the -free route stops; later prices, cache rates and exact cutoff time are unverified.", "documentedRates": { "input": 0, "output": 0, "cached": null, "label": "Boundless Gateway promotion \u00b7 USD" }, "accessNote": "Use convaiinnovations/laya or convaiinnovations/laya-free. The published zero quote is temporary and provider-specific.", "billingNote": "Temporary Boundless promotion; later rates and fees are unverified.", "checked": "2026-10-09", "catalogId": null, "catalogAdded": null, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "rates": { "input": null, "output": null, "cached": null, "write": null, "overrides": [] }, "independentEvidence": [] }, { "slug": "gemini-4-argon", "providerId": null, "name": "Gemini 4 Argon", "provider": "Google", "verification": "Announced \u00b7 restricted access", "discovered": "2026-10-03", "released": null, "announced": "2026-09-30", "context": null, "maxOutput": 1000000, "modalities": [ "text", "image", "video" ], "tools": null, "sourceUrl": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-4-argon/", "note": "Google announced Argon for phased access through the Fairwind Program. Broad availability and a public API model ID remain unverified.", "capabilityNote": "Google announces multimodal capabilities and a 1M-token output allowance. This is an output limit, not a verified input or combined context window.", "maxOutputNote": "The announced output allowance does not establish an endpoint contract or measured output length.", "pricingNote": "Announced introductory rates are $2 input and $10 output per million tokens, with 95% discounted cached input ($0.10). Later announced rates are $4/$20. The global post gives no exact expiry date; these are future launch prices, not verified public API billing.", "documentedRates": { "input": 2, "output": 10, "cached": 0.1, "label": "Announced introductory rates \u00b7 USD" }, "accessNote": "Initial access is restricted. Independent Arena High preference evidence appears below; matched LiveBench and effort sweeps remain unavailable.", "checked": "2026-10-09", "catalogId": null, "catalogAdded": null, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "rates": { "input": null, "output": null, "cached": null, "write": null, "overrides": [] }, "independentEvidence": [ { "source": "Arena text preference", "modelId": "gemini-4-argon-high", "score": 1525, "uncertainty": 9, "votes": 4892, "preliminary": true, "sourceDate": "2026-10-08", "checked": "2026-10-09", "sourceUrl": "https://arena.ai/leaderboard/text", "note": "Preference evidence for this exact configuration. It is not a percent-correct score or an effort sweep. This row is not merged with a differently configured LiveBench result." } ] }, { "slug": "cloudflare-clef-flash", "providerId": "@cf/cloudflare/clef-flash", "name": "Cloudflare Clef Flash", "provider": "Cloudflare", "discovered": "2026-10-02", "released": "2026-10-01", "context": 65536, "contextNote": "Hosted Workers AI context is 65,536 tokens. API documentation warns that long text state is truncated to fit the limit; retain a margin for questions and inspect truncation in your workload.", "maxOutput": null, "modalities": [ "text", "image", "video" ], "tools": false, "specialized": "decision", "questionLimit": 64, "outputType": "Yes/no probabilities, bounded choices and rubric scores", "sourceUrl": "https://developers.cloudflare.com/workers-ai/models/clef-flash/", "officialUrl": "https://blog.cloudflare.com/clef-decision-models/", "note": "Provider-announced multimodal decision model with Apache 2.0 weights. Workers AI exposes bounded typed questions, not a matched chat-model result. Vendor quality/latency comparisons have not been independently reproduced here.", "capabilityNote": "The model card declares text, JSON, images and video input. The hosted schema explicitly documents 1\u201364 named questions and up to four embedded images; hosted video request support is not established by the model-card claim. No adjustable effort sweep is verified.", "pricingNote": "Workers AI lists $0.09 per million input tokens for this exact endpoint. Its table lists no separate output rate; output and cache rates remain unverified here. Check actual usage and billing, including question overhead, before comparing task costs.", "documentedRates": { "input": 0.09, "output": null, "cached": null, "label": "Workers AI input quote \u00b7 USD" }, "accessNote": "Hosted through Workers AI using a Cloudflare account and API authorization; Apache 2.0 weights can be run separately. Hosting costs and local inference costs differ.", "checked": "2026-10-09", "catalogId": null, "catalogAdded": null, "verification": "Provider documented", "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "rates": { "input": null, "output": null, "cached": null, "write": null, "overrides": [] }, "independentEvidence": [] }, { "slug": "cloudflare-clef", "providerId": "@cf/cloudflare/clef", "name": "Cloudflare Clef", "provider": "Cloudflare", "discovered": "2026-10-02", "released": "2026-10-01", "context": 65536, "contextNote": "Hosted Workers AI context is 65,536 tokens. API documentation warns that long text state is truncated to fit the limit; retain a margin for questions and inspect truncation in your workload.", "maxOutput": null, "modalities": [ "text", "image", "video" ], "tools": false, "specialized": "decision", "questionLimit": 64, "outputType": "Yes/no probabilities, bounded choices and rubric scores", "sourceUrl": "https://developers.cloudflare.com/workers-ai/models/clef/", "officialUrl": "https://blog.cloudflare.com/clef-decision-models/", "note": "Provider-announced multimodal decision model with Apache 2.0 weights. Workers AI exposes bounded typed questions, not a matched chat-model result. Vendor quality/latency comparisons have not been independently reproduced here.", "capabilityNote": "The model card declares text, JSON, images and video input. The hosted schema explicitly documents 1\u201364 named questions and up to four embedded images; hosted video request support is not established by the model-card claim. No adjustable effort sweep is verified.", "pricingNote": "Workers AI lists $0.24 per million input tokens for this exact endpoint. Its table lists no separate output rate; output and cache rates remain unverified here. Check actual usage and billing, including question overhead, before comparing task costs.", "documentedRates": { "input": 0.24, "output": null, "cached": null, "label": "Workers AI input quote \u00b7 USD" }, "accessNote": "Hosted through Workers AI using a Cloudflare account and API authorization; Apache 2.0 weights can be run separately. Hosting costs and local inference costs differ.", "checked": "2026-10-09", "catalogId": null, "catalogAdded": null, "verification": "Provider documented", "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "rates": { "input": null, "output": null, "cached": null, "write": null, "overrides": [] }, "independentEvidence": [] }, { "slug": "perplexity-decider-v1-27b", "providerId": "pplx-decider-v1-27b", "name": "Perplexity Decider v1 27B", "provider": "Perplexity", "discovered": "2026-10-02", "released": null, "context": 262144, "contextNote": "The Decisions API accepts fewer than 262,144 total input tokens, including state, images and all questions; this is an input cap, not a chat prompt-plus-output context allowance.", "maxOutput": null, "modalities": [ "text", "image" ], "tools": false, "specialized": "decision", "questionLimit": 128, "outputType": "Yes/no probabilities, bounded choices and rubric scores", "sourceUrl": "https://docs.perplexity.ai/docs/decisions/quickstart", "officialUrl": "https://huggingface.co/perplexity-ai/pplx-decider-v1-27b", "note": "Multimodal decision model with Apache 2.0 weights and a dedicated Decisions API. It returns bounded probabilities rather than open-ended chat or code. Vendor-reported evaluations are separate from independent matched evidence; the provider launch date remains unverified.", "capabilityNote": "The dedicated Decisions API accepts text, JSON and images and supports 1\u2013128 named questions in a request. It returns typed probabilities; no chat tool-calling interface or adjustable reasoning-effort sweep is established.", "pricingNote": "Direct Decisions API billing is $0.04 per million response-reported input tokens, with free output and no per-request fee. State, all questions and images count toward input; caching is unverified.", "documentedRates": { "input": 0.04, "output": 0, "cached": null, "label": "Direct Decisions API \u00b7 USD" }, "accessNote": "Public documentation accepts any Perplexity API key. This review made no private-key inference request.", "checked": "2026-10-09", "catalogId": null, "catalogAdded": null, "verification": "Provider documented", "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "rates": { "input": null, "output": null, "cached": null, "write": null, "overrides": [] }, "independentEvidence": [] }, { "slug": "ling-3-1-flash", "providerId": "inclusionai/ling-3.1-flash", "name": "Ling 3.1 Flash", "provider": "InclusionAI", "verification": "Hosted route documented", "identityLabel": "AI Gateway route", "discovered": "2026-10-02", "released": null, "context": 262144, "contextNote": "AI Gateway\u2019s exact models API declares a 262,144-token hosted context and 32,768-token output allowance; the model webpage rounds context to 262K. Gateway availability was announced September 30, separately from the unverified base-model launch date.", "maxOutput": 32768, "modalities": [ "text" ], "tools": true, "sourceUrl": "https://vercel.com/ai-gateway/models/ling-3.1-flash", "officialUrl": "https://vercel.com/changelog/ling-3-1-flash-is-now-available-on-ai-gateway", "note": "Hybrid reasoning model with a temporary Vercel AI Gateway promotion through October 13, 2026. The standard route begins billing afterward; the -free route stops serving. Exact cutoff time and later rates are unverified. No independent matched score is assigned.", "accessNote": "The alternate inclusionai/ling-3.1-flash-free route stops serving at promotion expiry instead of beginning paid billing. This promotion is specific to AI Gateway; it does not establish a free direct-provider API.", "pricingNote": "The cited Gateway promotion currently has free input and output. Post-promotion prices and exact expiry time are unverified; no permanent zero rate or routed OpenRouter offer is inferred.", "documentedRates": { "input": 0, "output": 0, "cached": null, "label": "AI Gateway promotion \u00b7 USD" }, "capabilityNote": "Vercel documents text input, tool use, hybrid reasoning and a 32,768-token output cap including reasoning. Its exact Gateway catalog lists a reasoning toggle plus none/low/medium/high effort controls; a native mapping and default are unverified, and no matched independent effort sweep is available.", "gatewayEffortOptions": [ "none", "low", "medium", "high" ], "capabilitySourceUrl": "https://ai-gateway.vercel.sh/v1/models", "catalogOffer": { "id": "inclusionai/ling-3.1-flash", "sourceUrl": "https://openrouter.ai/inclusionai/ling-3.1-flash", "observed": "2026-10-04", "input": 0, "output": 0, "cached": null, "note": "Newly listed OpenRouter offer. Zero input/output are the observed routed quote; duration, capacity and native API defaults are unverified. This offer is separate from the Gateway promotion.", "checked": "2026-10-07" }, "checked": "2026-10-09", "catalogId": null, "catalogAdded": null, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "rates": { "input": null, "output": null, "cached": null, "write": null, "overrides": [] }, "independentEvidence": [] }, { "slug": "minimax-m3-1-flash-preview", "providerId": "MiniMax-M3.1-Flash-Preview", "name": "MiniMax M3.1 Flash Preview", "provider": "MiniMax", "discovered": "2026-10-01", "context": 1000000, "maxOutput": null, "modalities": [ "text", "image", "video" ], "tools": true, "supportedEfforts": [ "low", "medium", "high", "xhigh", "max" ], "sourceUrl": "https://www.minimax.io/m-plan", "officialUrl": "https://agent.minimax.io/tools/m3-1-flash-preview", "note": "Provider-documented preview with always-on reasoning and five effort controls; omitted API effort defaults to max. Currently available through M Plan and MiniMax Code. Launch date, public per-token pricing, measured speed and exact matched independent benchmarks remain unverified.", "defaultEffort": "max", "apiDocumentationUrls": [ "https://platform.minimax.io/docs/api-reference/text-openai-api", "https://platform.minimax.io/docs/api-reference/text-anthropic-api" ], "accessNote": "Provider API documentation currently limits this preview to M Plan and MiniMax Code. Verify your account and subscription-key entitlement; a standard pay-as-you-go key must not be assumed to have access.", "maxOutputNote": "A provider maximum output allowance is not verified. The 4,096-token request in the Anthropic documentation is an example, not a published model limit; thinking and answer tokens share the request budget.", "checked": "2026-10-09", "catalogId": null, "catalogAdded": null, "released": null, "verification": "Provider documented", "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "rates": { "input": null, "output": null, "cached": null, "write": null, "overrides": [] }, "independentEvidence": [] }, { "slug": "ling-3-0-flash-sante", "catalogId": "inclusionai/ling-3.0-flash-sante", "name": "Ling 3.0 Flash Sante", "provider": "InclusionAI", "released": null, "catalogAdded": "2026-09-04", "checked": "2026-10-09", "verification": "Paid catalog offer", "note": "The paid route replaces a formerly free catalog entry for this health-focused text model. This is a serving-offer change, not a verified new base-model launch. Medical-domain claims and tool support are catalog declarations; no independent clinical reliability is established.", "context": 262144, "maxOutput": 32768, "modalities": [ "text" ], "tools": true, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "sourceUrl": "https://openrouter.ai/inclusionai/ling-3.0-flash-sante", "officialUrl": null, "rates": { "input": 0.042, "output": 0.1232, "cached": 0.0084, "write": null, "overrides": [] }, "minimumCacheTokens": null, "independentEvidence": [] }, { "slug": "step-5-preview", "catalogId": "stepfun/step-5-preview", "name": "Step 5 Preview", "provider": "StepFun", "released": null, "catalogAdded": "2026-10-08", "checked": "2026-10-09", "verification": "Provider documented preview", "note": "A new routed offer for StepFun\u2019s coding and knowledge-work model. Native documentation supports text, image and video input, tools, 1M context and 64K output. The Gateway output allowance differs; check your exact endpoint. No matched independent score is assigned.", "context": 1000000, "maxOutput": 64000, "modalities": [ "text", "image", "video" ], "tools": true, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "sourceUrl": "https://openrouter.ai/stepfun/step-5-preview", "officialUrl": "https://platform.stepfun.com/docs/zh/guides/models/step-5-preview", "rates": { "input": 1.0, "output": 2.7, "cached": 0.05, "write": null, "overrides": [] }, "minimumCacheTokens": null, "supportedEfforts": [ "low", "medium", "high" ], "defaultEffort": null, "discovered": "2026-10-09", "dateNote": "OpenRouter addition is October 8. Gateway has an older release-metadata date; neither is treated as a verified native launch date.", "contextNote": "Native and OpenRouter documentation agree on 1M context and 64K output. Gateway currently lists 1M max_tokens and text/image input; that serving declaration does not override the native 64K cap or establish video support there.", "capabilityNote": "Native API ID step-5-preview accepts text, images and video; supports tools, JSON/JSON Schema and low/medium/high effort. The application provides tools and external access. The routed offer defaults to medium, but the native default is not explicitly established by the reviewed reference.", "pricingNote": "OpenRouter quotes $1 input, $0.05 cached input and $2.70 output per million, observed October 9. Native billing is separately denominated in CNY: \u00a57 input, \u00a50.35 cache reads and \u00a520 output per million. Thinking and final text both count as output; no currency conversion is substituted for either offer.", "independentEvidence": [] }, { "slug": "claude-haiku-5-5", "catalogId": "anthropic/claude-haiku-5.5", "name": "Claude Haiku 5.5", "provider": "Anthropic", "released": "2026-10-07", "catalogAdded": "2026-10-07", "checked": "2026-10-09", "verification": "Provider verified", "note": "Five native effort levels with medium default; thinking is on by default. The 1M context and 128K output limits are documented. Prompts above 100K use five times the base token rates. Its newer tokenizer can count the same text differently from Haiku 4.5; compare actual billed usage before switching.", "context": 1000000, "maxOutput": 128000, "modalities": [ "text", "image", "file" ], "tools": true, "benchmarkStatus": "Matched LiveBench", "benchmarkSlug": "claude-haiku-5-5", "sourceUrl": "https://openrouter.ai/anthropic/claude-haiku-5.5", "officialUrl": "https://platform.claude.com/docs/en/models/haiku-5-5/overview", "rates": { "input": 0.1, "output": 0.5, "cached": 0.01, "write": 0.125, "overrides": [ { "min_prompt_tokens": 100000, "prompt": 0.5, "completion": 2.5, "input_cache_read": 0.05, "input_cache_write": 0.625, "input_cache_write_1h": "0.000001" } ] }, "minimumCacheTokens": 512, "providerId": "claude-haiku-5-5", "supportedEfforts": [ "low", "medium", "high", "xhigh", "max" ], "defaultEffort": "medium", "thinkingDefault": "adaptive", "capabilityNote": "Thinking can be disabled at low, medium or high effort, but that request is rejected at xhigh/max. Manual budget_tokens thinking is unsupported. Omit sampling parameters and handle thinking blocks by type. The newer tokenizer counts about 30% more tokens for the same text on average; actual usage varies.", "priceContract": { "threshold": 100000, "comparison": "strictly greater", "base": { "input": 0.1, "output": 0.5, "cached": 0.01, "write": 0.125, "write1h": 0.2 }, "long": { "input": 0.5, "output": 2.5, "cached": 0.05, "write": 0.625, "write1h": 1 }, "sourceUrl": "https://platform.claude.com/docs/en/about-claude/pricing" }, "independentEvidence": [] }, { "slug": "mistral-large-4-0", "catalogId": "mistralai/mistral-large-4-0", "name": "Mistral Large 4", "provider": "Mistral", "released": "2026-10-06", "catalogAdded": "2026-10-06", "checked": "2026-10-09", "verification": "Public preview", "note": "Preview API access is available; open weights are planned for later in October. The October 9 routed offer now declares 1,048,576 context tokens, up from 524,288, and 262,144 maximum output. The two-week 50% launch discount announced October 6 is unchanged; exact cutoff time is unverified.", "context": 1048576, "maxOutput": 262144, "modalities": [ "text", "image" ], "tools": true, "benchmarkStatus": "Matched LiveBench", "benchmarkSlug": "mistral-large-4-0", "sourceUrl": "https://openrouter.ai/mistralai/mistral-large-4-0", "officialUrl": "https://docs.mistral.ai/models/mistral-large-4-0", "rates": { "input": 0.68, "output": 2.09, "cached": 0.07, "write": null, "overrides": [] }, "minimumCacheTokens": null, "independentEvidence": [] }, { "slug": "switchyard", "catalogId": "nvidia/switchyard", "name": "NVIDIA Switchyard", "provider": "NVIDIA", "released": null, "catalogAdded": "2026-09-21", "checked": "2026-10-09", "verification": "Routing system", "note": "A router over a configured model pool, not a new base model. OpenRouter now explicitly states that each request uses its serving model\u2019s price and context allowance. Negative catalog price sentinels remain unknown fixed rates; no free service or guaranteed one-million-token request is inferred.", "context": 1000000, "maxOutput": null, "modalities": [ "text" ], "tools": false, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "sourceUrl": "https://openrouter.ai/nvidia/switchyard", "officialUrl": "https://github.com/NVIDIA-NeMo/Switchyard", "rates": { "input": null, "output": null, "cached": null, "write": null, "overrides": [] }, "minimumCacheTokens": null, "independentEvidence": [] }, { "slug": "pareto-26-10-preview", "catalogId": "unbiased/pareto-26.10-preview", "name": "Pareto 26.10 Preview", "provider": "Unbiased", "released": "2026-10-01", "catalogAdded": "2026-10-01", "checked": "2026-10-09", "verification": "Composite system", "note": "Provider-documented blended system, not an independent base-model checkpoint. Preview serving behavior can change. Vendor preliminary results use different question sets and harnesses; no matched score is inferred here.", "context": 1048576, "maxOutput": 131072, "modalities": [ "text", "image" ], "tools": true, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "sourceUrl": "https://openrouter.ai/unbiased/pareto-26.10-preview", "officialUrl": "https://unbiased.ai/blog/pareto-26-10-preview/", "rates": { "input": 0.8, "output": 3.2, "cached": 0.03, "write": null, "overrides": [] }, "minimumCacheTokens": null, "independentEvidence": [] }, { "slug": "apodex-1-1-mini-free", "catalogId": "apodex/apodex-1.1-mini:free", "name": "Apodex 1.1 Mini (free route)", "provider": "Apodex", "released": null, "catalogAdded": "2026-10-01", "checked": "2026-10-09", "verification": "Catalog offer", "note": "New free catalog route for an existing model, not a verified new base-model launch. Direct Apodex rates are separately documented at $0.08 input, $0.008 cache read and $0.80 output per million tokens; they must not be substituted for this zero-priced route. Exact independent results and free-route capacity remain unverified.", "context": 262144, "maxOutput": 235929, "modalities": [ "text" ], "tools": true, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "sourceUrl": "https://openrouter.ai/apodex/apodex-1.1-mini:free", "officialUrl": "https://platform.apodex.ai/docs/models", "rates": { "input": 0.0, "output": 0.0, "cached": null, "write": null, "overrides": [] }, "minimumCacheTokens": null, "independentEvidence": [] }, { "slug": "gpt-6-1-sol", "catalogId": "openai/gpt-6.1-sol", "name": "GPT-6.1 Sol", "provider": "OpenAI", "released": "2026-09-29", "catalogAdded": "2026-09-29", "checked": "2026-10-09", "verification": "Provider verified", "note": "Five supported effort levels: low, medium (the API default), high, extra high and max. None and minimal are unsupported. Tool calling uses Responses; Chat Completions supports requests without tools.", "context": 1050000, "maxOutput": 128000, "modalities": [ "file", "image", "text" ], "tools": true, "benchmarkStatus": "Matched LiveBench", "benchmarkSlug": "gpt-6-1-sol", "sourceUrl": "https://openrouter.ai/openai/gpt-6.1-sol", "officialUrl": "https://developers.openai.com/api/docs/models/gpt-6.1-sol", "rates": { "input": 2.0, "output": 10.0, "cached": 0.1, "write": 2.5, "overrides": [ { "min_prompt_tokens": 272000, "prompt": 4.0, "completion": 15.0, "input_cache_read": 0.2, "input_cache_write": 5.0 } ] }, "minimumCacheTokens": null, "independentEvidence": [] }, { "slug": "claude-sonnet-5-5", "catalogId": "anthropic/claude-sonnet-5.5", "name": "Claude Sonnet 5.5", "provider": "Anthropic", "released": "2026-09-28", "catalogAdded": "2026-09-28", "checked": "2026-10-09", "verification": "Provider verified", "note": "Adaptive thinking, high default effort and 128K maximum output. Migration changes affect forced tool use, thinking replay and older computer-use tools.", "context": 1000000, "maxOutput": 128000, "modalities": [ "text", "image", "file" ], "tools": true, "benchmarkStatus": "Matched LiveBench", "benchmarkSlug": "claude-sonnet-5-5", "sourceUrl": "https://openrouter.ai/anthropic/claude-sonnet-5.5", "officialUrl": "https://platform.claude.com/docs/en/models/sonnet-5-5/overview", "rates": { "input": 2.0, "output": 10.0, "cached": 0.1, "write": 2.5, "overrides": [] }, "minimumCacheTokens": 512, "independentEvidence": [ { "source": "Arena text preference", "modelId": "claude-sonnet-5.5-xhigh", "score": 1476, "uncertainty": 9, "votes": 4838, "preliminary": false, "sourceDate": "2026-10-08", "checked": "2026-10-09", "sourceUrl": "https://arena.ai/leaderboard/text", "note": "Preference evidence for this exact configuration. It is not a percent-correct score or an effort sweep. This row is not merged with a differently configured LiveBench result." } ] }, { "slug": "gpt-6-1-sol-pro", "catalogId": "openai/gpt-6.1-sol-pro", "name": "GPT-6.1 Sol Pro", "provider": "OpenAI", "released": null, "catalogAdded": "2026-09-29", "checked": "2026-10-09", "verification": "Catalog variant", "note": "OpenRouter describes this as GPT-6.1 Sol served in pro mode, with greater token use. It is a serving variant; no independent result is assigned from the base model.", "context": 1050000, "maxOutput": 128000, "modalities": [ "file", "image", "text" ], "tools": true, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "sourceUrl": "https://openrouter.ai/openai/gpt-6.1-sol-pro", "officialUrl": null, "rates": { "input": 2.0, "output": 10.0, "cached": 0.1, "write": 2.5, "overrides": [ { "min_prompt_tokens": 272000, "prompt": 4.0, "completion": 15.0, "input_cache_read": 0.2, "input_cache_write": 5.0 } ] }, "minimumCacheTokens": null, "independentEvidence": [] }, { "slug": "glm-5-3-prime", "catalogId": "z-ai/glm-5.3-prime", "name": "GLM 5.3 Prime", "provider": "Z.ai", "released": null, "catalogAdded": "2026-09-23", "checked": "2026-10-09", "verification": "Catalog variant", "note": "Separate catalog endpoint. A throughput claim does not establish measured speed here; base-model benchmark scores are withheld.", "context": 1000000, "maxOutput": 131072, "modalities": [ "text" ], "tools": true, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "sourceUrl": "https://openrouter.ai/z-ai/glm-5.3-prime", "officialUrl": null, "rates": { "input": 2.8, "output": 8.8, "cached": 0.56, "write": null, "overrides": [] }, "minimumCacheTokens": null, "independentEvidence": [] }, { "slug": "qwen3-8-max-prime", "catalogId": "qwen/qwen3.8-max-prime", "name": "Qwen3.8 Max Prime", "provider": "Alibaba", "released": null, "catalogAdded": "2026-09-23", "checked": "2026-10-09", "verification": "Catalog variant", "note": "Separate catalog endpoint. No exact matched benchmark result or measured speed is verified for this variant.", "context": 1000000, "maxOutput": 131072, "modalities": [ "text", "image", "video" ], "tools": true, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "sourceUrl": "https://openrouter.ai/qwen/qwen3.8-max-prime", "officialUrl": null, "rates": { "input": 4.0, "output": 12.0, "cached": 0.5, "write": null, "overrides": [] }, "minimumCacheTokens": null, "independentEvidence": [] }, { "slug": "ember-1", "catalogId": "fireworks/ember-1", "name": "Ember-1", "provider": "Fireworks", "released": "2026-09-23", "catalogAdded": "2026-09-24", "checked": "2026-10-09", "verification": "Provider verified", "note": "Research preview announced September 23 with temporary two-week serverless access. Vendor claims about shorter reasoning are not independently measured task costs; verify availability.", "context": 1048576, "maxOutput": 943718, "modalities": [ "text", "image" ], "tools": true, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "sourceUrl": "https://openrouter.ai/fireworks/ember-1", "officialUrl": "https://fireworks.ai/blog/ember-1", "rates": { "input": 3.0, "output": 15.0, "cached": 0.3, "write": null, "overrides": [] }, "minimumCacheTokens": null, "independentEvidence": [] }, { "slug": "aion-3-5", "catalogId": "aion-labs/aion-3.5", "name": "Aion 3.5", "provider": "AionLabs", "released": null, "catalogAdded": "2026-09-23", "checked": "2026-10-09", "verification": "Catalog verified", "note": "Catalog-listed text generation system. Independent general-task or storytelling performance has not been verified in our matched datasets.", "context": 262144, "maxOutput": 32768, "modalities": [ "text" ], "tools": true, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "sourceUrl": "https://openrouter.ai/aion-labs/aion-3.5", "officialUrl": null, "rates": { "input": 3.0, "output": 6.0, "cached": 0.75, "write": null, "overrides": [] }, "minimumCacheTokens": null, "independentEvidence": [] }, { "slug": "aion-3-5-mini", "catalogId": "aion-labs/aion-3.5-mini", "name": "Aion 3.5 Mini", "provider": "AionLabs", "released": null, "catalogAdded": "2026-09-23", "checked": "2026-10-09", "verification": "Catalog verified", "note": "Lower-priced text generation endpoint. A lower price does not establish equal quality or cost per successful task.", "context": 262144, "maxOutput": 32768, "modalities": [ "text" ], "tools": true, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "sourceUrl": "https://openrouter.ai/aion-labs/aion-3.5-mini", "officialUrl": null, "rates": { "input": 0.7, "output": 1.4, "cached": 0.18, "write": null, "overrides": [] }, "minimumCacheTokens": null, "independentEvidence": [] }, { "slug": "perceptron-mk1-5", "catalogId": "perceptron/perceptron-mk1.5", "name": "Perceptron Mk1.5", "provider": "Perceptron", "released": "2026-09-25", "catalogAdded": "2026-09-25", "checked": "2026-10-09", "verification": "Provider verified", "note": "Provider verifies text, image, audio and video input and 32K context; this catalog offer advertises 36,864 tokens. Confirm the selected endpoint limit. Specialized performance is not scored using another model\u2019s text benchmarks.", "context": 36864, "maxOutput": 8192, "modalities": [ "text", "image", "video", "audio" ], "tools": true, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "sourceUrl": "https://openrouter.ai/perceptron/perceptron-mk1.5", "officialUrl": "https://www.perceptron.inc/blog/introducing-perceptron-mk1-5", "rates": { "input": 0.15, "output": 1.5, "cached": null, "write": null, "overrides": [] }, "minimumCacheTokens": null, "independentEvidence": [] } ], "baselines": [ { "slug": "gpt-6-sol", "catalogId": "openai/gpt-6-sol", "name": "GPT-6 Sol", "provider": "OpenAI", "released": null, "catalogAdded": "2026-09-22", "checked": "2026-10-09", "verification": "Catalog verified", "note": "", "context": 1050000, "maxOutput": 128000, "modalities": [ "file", "image", "text" ], "tools": true, "benchmarkStatus": "Matched LiveBench", "benchmarkSlug": "gpt-6-sol", "sourceUrl": "https://openrouter.ai/openai/gpt-6-sol", "officialUrl": null, "rates": { "input": 2.0, "output": 10.0, "cached": 0.2, "write": 2.5, "overrides": [ { "min_prompt_tokens": 272000, "prompt": 4.0, "completion": 15.0, "input_cache_read": 0.4, "input_cache_write": 5.0 } ] }, "minimumCacheTokens": null }, { "slug": "claude-haiku-4-5", "catalogId": "anthropic/claude-haiku-4.5", "name": "Claude Haiku 4.5", "provider": "Anthropic", "released": null, "catalogAdded": "2025-10-15", "checked": "2026-10-09", "verification": "Catalog verified", "note": "", "context": 200000, "maxOutput": 64000, "modalities": [ "text", "image", "file" ], "tools": true, "benchmarkStatus": "Matched LiveBench pending", "benchmarkSlug": null, "sourceUrl": "https://openrouter.ai/anthropic/claude-haiku-4.5", "officialUrl": null, "rates": { "input": 1.0, "output": 5.0, "cached": 0.1, "write": 1.25, "overrides": [] }, "minimumCacheTokens": 4096 }, { "slug": "gpt-6-astra", "catalogId": "openai/gpt-6-astra", "name": "GPT-6 Astra", "provider": "OpenAI", "released": null, "catalogAdded": "2026-09-04", "checked": "2026-10-09", "verification": "Catalog verified", "note": "", "context": 1050000, "maxOutput": 128000, "modalities": [ "file", "image", "text" ], "tools": true, "benchmarkStatus": "Matched LiveBench", "benchmarkSlug": "gpt-6-astra", "sourceUrl": "https://openrouter.ai/openai/gpt-6-astra", "officialUrl": null, "rates": { "input": 10.0, "output": 50.0, "cached": 1.0, "write": 12.5, "overrides": [ { "min_prompt_tokens": 272000, "prompt": 20.0, "completion": 75.0, "input_cache_read": 2.0, "input_cache_write": 25.0 } ] }, "minimumCacheTokens": null }, { "slug": "claude-sonnet-5", "catalogId": "anthropic/claude-sonnet-5", "name": "Claude Sonnet 5", "provider": "Anthropic", "released": null, "catalogAdded": "2026-06-30", "checked": "2026-10-09", "verification": "Catalog verified", "note": "", "context": 1000000, "maxOutput": 128000, "modalities": [ "text", "image", "file" ], "tools": true, "benchmarkStatus": "Matched LiveBench", "benchmarkSlug": "claude-sonnet-5", "sourceUrl": "https://openrouter.ai/anthropic/claude-sonnet-5", "officialUrl": null, "rates": { "input": 2.0, "output": 10.0, "cached": 0.2, "write": 2.5, "overrides": [] }, "minimumCacheTokens": null }, { "slug": "claude-opus-5-5", "catalogId": "anthropic/claude-opus-5.5", "name": "Claude Opus 5.5", "provider": "Anthropic", "released": null, "catalogAdded": "2026-09-22", "checked": "2026-10-09", "verification": "Catalog verified", "note": "", "context": 1000000, "maxOutput": 128000, "modalities": [ "text", "image", "file" ], "tools": true, "benchmarkStatus": "Matched LiveBench", "benchmarkSlug": "claude-opus-5-5", "sourceUrl": "https://openrouter.ai/anthropic/claude-opus-5.5", "officialUrl": null, "rates": { "input": 4.0, "output": 20.0, "cached": 0.2, "write": 5.0, "overrides": [] }, "minimumCacheTokens": null } ], "checks": [ { "source": "LiveBench", "status": "June 25 question set, all 69 exact rows and category definitions are byte-identical to October 8. No new matched effort sweep.", "url": "https://livebench.ai/table_2026_06_25.csv" }, { "source": "Arena", "status": "October 8 source observation refreshes twelve exact preference rows, including score/uncertainty/votes. Ten exact matches enter paired evidence; Argon high and Sonnet xhigh remain separately scoped.", "url": "https://arena.ai/leaderboard/text" }, { "source": "Hex DataBench", "status": "Old 64-row dataset and September22 date preserved. Publisher methodology link checked; no restricted numeric acquisition or new import.", "url": "https://hex.tech/databench/" }, { "source": "OckBench", "status": "Both exact numeric datasets unchanged; prior task/settings/source dates retained.", "url": "https://ockbench.github.io/static/data/top200_model_performance.csv" }, { "source": "OpenRouter", "status": "469 routes, four additions and two removals. Step 5 Preview and paid Ling Sante are new covered offers; two Sol batch SKUs are serving modes. 7 covered dated quotes changed, including an unchanged Pro0813 UTC schedule.", "url": "https://openrouter.ai/api/v1/models" }, { "source": "OpenAI", "status": "October 8 Sol Ultrafast support and native short/long/residency prices verified. It is a service tier, not a new base model or effort result.", "url": "https://developers.openai.com/api/docs/changelog" }, { "source": "Anthropic", "status": "Existing Haiku/Sonnet identities, cache pricing, effort and migration rules rechecked. The agent web_fetch URL guard update is operational, not a model release.", "url": "https://platform.claude.com/docs/en/release-notes/overview" }, { "source": "Google", "status": "October 8 Deep Research December 2025 agent deprecation verified for October 23, with April 2026 agent replacements. No new general text-model release in checked sources.", "url": "https://ai.google.dev/gemini-api/docs/changelog" }, { "source": "DeepSeek", "status": "Native release notes and pricing still describe the existing family. Two Flash routed input quotes fell; Pro 0813 still uses its recorded UTC/day schedule.", "url": "https://api-docs.deepseek.com/quick_start/pricing/" }, { "source": "xAI", "status": "Grok4.7 remains the latest verified text model in checked native notes. October2 speech endpoint redirect is outside the text ranking.", "url": "https://docs.x.ai/developers/release-notes" }, { "source": "Alibaba/Qwen", "status": "Publisher checkpoint inventory adds no exact new ID. Qwen3.6 27B routed input/output rose and the cache quote is now absent.", "url": "https://huggingface.co/Qwen" }, { "source": "Z.ai", "status": "No newer native base-model ID found. Routed GLM5.2 input/output fell; GLM5.3 input fell while output rose.", "url": "https://docs.z.ai/release-notes/new-released" }, { "source": "Moonshot", "status": "Current KimiK3 routed input, output and cache prices all fell; no newly verified native model version.", "url": "https://www.moonshot.ai/" }, { "source": "MiniMax", "status": "M3.1 Flash remains a restricted preview. Example output budgets still do not establish a maximum output cap.", "url": "https://agent.minimax.io/tools/m3-1-flash-preview" }, { "source": "Meta", "status": "Existing MuseSpark contracts remain. Arena\u2019s xHigh1.2 observation changed; it is not mapped to another effort or version.", "url": "https://dev.meta.ai/docs/models" }, { "source": "NVIDIA", "status": "Switchyard remains a router with variable model billing. Current native announcement and model inventory establish no new covered base-model ID.", "url": "https://developer.nvidia.com/topics/ai/nemotron" }, { "source": "Thinking Machines", "status": "Inkling remains in the current Tinker inventory. No newer exact checkpoint or matched effort evidence found in the checked news/model sources.", "url": "https://tinker-docs.thinkingmachines.ai/tinker/models/" }, { "source": "BFCL", "status": "Exact current CSV unchanged. Reference remains separate from general quality ratings.", "url": "https://gorilla.cs.berkeley.edu/data_overall.csv" }, { "source": "Vectara", "status": "Exact README unchanged; actual source observation remains September22.", "url": "https://github.com/vectara/hallucination-leaderboard" }, { "source": "Fireworks", "status": "Ember1 remains a dated preview; its introductory access window is not treated as permanent pricing or proof of continued access.", "url": "https://fireworks.ai/blog/ember-1" }, { "source": "Perceptron", "status": "Mk1.5 multimodal specifications remain. A routed context allowance still does not override the provider\u2019s own documented limit.", "url": "https://www.perceptron.inc/blog/introducing-perceptron-mk1-5" }, { "source": "InclusionAI / Vercel", "status": "Ling3.1 temporary free Gateway offer is separate from the newly paid Ling3.0 FlashSante route. The October13 promotional cutoff remains.", "url": "https://vercel.com/changelog/ling-3-1-flash-is-now-available-on-ai-gateway" }, { "source": "Apodex", "status": "The reviewed free route and native paid offer remain separate; no new version or matched independent row found.", "url": "https://platform.apodex.ai/docs/models" }, { "source": "Unbiased", "status": "Pareto is still a blended preview system. Its components\u2019 scores are not assigned as this system\u2019s measured quality.", "url": "https://unbiased.ai/blog/pareto-26-10-preview/" }, { "source": "Perplexity", "status": "Decider exact checkpoint SHA is unchanged. Decision API limits and free-output contract remain separately scoped.", "url": "https://docs.perplexity.ai/docs/decisions/quickstart" }, { "source": "Cloudflare", "status": "Clef and ClefFlash remain distinct decision endpoints. Hosted schema support is not expanded from model-card modality claims.", "url": "https://blog.cloudflare.com/clef-decision-models/" }, { "source": "Historical matched effort studies", "status": "OckBench data is unchanged. Stet and Hampson studies keep their original tasks, settings, sample sizes and dates; no newer matched sweep was verified.", "url": "https://www.stet.sh/blog/gpt-55-codex-graphql-reasoning-curve" }, { "source": "Vercel AI Gateway exact model catalog", "url": "https://ai-gateway.vercel.sh/v1/models", "status": "413 routes: Apex is new, V4.1 FlashFast is absent. Step5 native/Gateway output and tool declarations differ and are kept explicit." }, { "source": "Convai Innovations / Vercel", "status": "Laya\u2019s temporary Boundless quote remains free through October31. Its unverified question cap stays unknown.", "url": "https://vercel.com/changelog/laya-decision-model-now-available-on-ai-gateway-free-through-october-31" }, { "source": "Artificial Analysis", "status": "Commercial reuse restrictions remain unresolved. No restricted leaderboard acquisition or new numeric import; historical values retain original dates.", "url": "https://artificialanalysis.ai/terms-of-use" }, { "source": "Mistral", "status": "Large4 routed context increased to1,048,576; native preview/promotion/weights status rechecked. Benchmark values unchanged.", "url": "https://docs.mistral.ai/resources/changelogs" }, { "source": "TypeSafe AI / Vercel", "status": "Jev\u2019s existing typed-answer contract remains. New Liquid d1 and local decision checkpoints add different access choices, rather than transferred Jev scores.", "url": "https://vercel.com/changelog/typesafe-ai-jev-now-available-on-ai-gateway" }, { "source": "Vercel production usage", "status": "Licensed export now covers September25\u2013October8:902 rows across fourteen complete UTC days and two populations. Missing named rows remain unreported.", "url": "https://vercel.com/docs/ai-gateway/leaderboards" }, { "source": "Reflection", "url": "https://developers.reflection.ai/models", "status": "Beam remains behind early-access enablement. Native effort/default/caps are unchanged; public token prices and promised weights remain unverified." }, { "source": "Reka", "url": "https://reka.ai/news/rho-1-collapsing-the-multimodal-stack", "status": "Rho1 remains an announced research system. The developer gateway does not establish ordinary Rho1 API access." }, { "source": "Aleph Alpha", "url": "https://huggingface.co/Aleph-Alpha/Kolibri-1", "status": "Kolibri exact checkpoint SHA is unchanged. Native262K and validated1M-extension claims remain separate from hosted limits." }, { "source": "Ai2", "url": "https://allenai.org/blog/astabrief", "status": "Exact AstaBrief checkpoint identity is unchanged. Ai2\u2019s October7 Bwen/Blama announcement and checkpoint contracts were newly reviewed separately." }, { "source": "Hex DataBench reuse permissions", "url": "https://learn.hex.tech/docs/legal/terms-and-conditions-2020-03-27", "status": "Published reuse restrictions remain unresolved. No new numeric imports or bulk leaderboard acquisition. Historical observations retain original dates." }, { "source": "StepFun", "url": "https://platform.stepfun.com/docs/zh/guides/models/step-5-preview", "status": "Native Step5 ID, modalities, tools, effort options, caps and CNY prices checked; no new independent row. Gateway maxoutput differs." }, { "source": "Callstack", "url": "https://apex.callstack.com/models", "status": "Pinned native Apex snapshot and October1 general availability verified; October8 Gateway offer separately scoped. Native price page404." }, { "source": "Liquid AI", "url": "https://www.liquid.ai/news/models", "status": "Hosted d1 and two October7 decision checkpoints verified with exact IDs, model-card limits and licenses. No independent matched decision evaluation imported." }, { "source": "Ai2 announcements", "url": "https://allenai.org/news", "status": "October7 Bwen/Blama byte-level research release verified; checkpoint dates, custom units and licenses distinguished. Stage1 training artifacts recorded without separate base-model counts." } ], "sourceHash": "6800ae8a4dcbcd12a02a954dc7ee170b37c3c784e096ba910e48adbe353d459c", "priceChanges": [ { "slug": "deepseek-v4-flash", "name": "DeepSeek V4 Flash", "catalogId": "deepseek/deepseek-v4-flash", "previousDate": "2026-10-08", "currentDate": "2026-10-09", "previous": { "input": 0.0228, "output": 1.28, "cached": 0.0228 }, "current": { "input": 0.0173, "output": 1.28, "cached": 0.0173 }, "scheduled": false, "previousScheduleRemoved": false, "note": "These are routed catalog observations, not a verified change to direct-provider list prices. Compare both input and output: one may fall while the other rises." }, { "slug": "glm-5-2", "name": "GLM-5.2", "catalogId": "z-ai/glm-5.2", "previousDate": "2026-10-08", "currentDate": "2026-10-09", "previous": { "input": 0.0855, "output": 12.0, "cached": 0.077 }, "current": { "input": 0.06, "output": 6.0, "cached": 0.059 }, "scheduled": false, "previousScheduleRemoved": false, "note": "These are routed catalog observations, not a verified change to direct-provider list prices. Compare both input and output: one may fall while the other rises." }, { "slug": "qwen3-6-27b", "name": "Qwen3.6 27B", "catalogId": "qwen/qwen3.6-27b", "previousDate": "2026-10-08", "currentDate": "2026-10-09", "previous": { "input": 0.3, "output": 2.0, "cached": 0.03 }, "current": { "input": 0.45, "output": 2.7, "cached": null }, "scheduled": false, "previousScheduleRemoved": false, "note": "These are routed catalog observations, not a verified change to direct-provider list prices. Compare both input and output: one may fall while the other rises. The current quote omits a cache-read rate; it is unknown, rather than free." }, { "slug": "kimi-k3", "name": "Kimi K3", "catalogId": "moonshotai/kimi-k3", "previousDate": "2026-10-08", "currentDate": "2026-10-09", "previous": { "input": 0.65, "output": 15.0, "cached": 0.43 }, "current": { "input": 0.5, "output": 12.0, "cached": 0.3 }, "scheduled": false, "previousScheduleRemoved": false, "note": "These are routed catalog observations, not a verified change to direct-provider list prices. Compare both input and output: one may fall while the other rises." }, { "slug": "deepseek-v4-flash-0731", "name": "DeepSeek V4 Flash 0731", "catalogId": "deepseek/deepseek-v4-flash-0731", "previousDate": "2026-10-08", "currentDate": "2026-10-09", "previous": { "input": 0.0137, "output": 1.28, "cached": 0.0137 }, "current": { "input": 0.0079, "output": 1.28, "cached": 0.0079 }, "scheduled": false, "previousScheduleRemoved": false, "note": "These are routed catalog observations, not a verified change to direct-provider list prices. Compare both input and output: one may fall while the other rises." }, { "slug": "deepseek-v4-pro-0813", "name": "DeepSeek V4 Pro 0813", "catalogId": "deepseek/deepseek-v4-pro-0813", "previousDate": "2026-10-08", "currentDate": "2026-10-09", "previous": { "input": 1.32, "output": 3.96, "cached": 0.044 }, "current": { "input": 0.66, "output": 1.98, "cached": 0.022 }, "scheduled": true, "previousScheduleRemoved": false, "note": "The UTC/day price schedule is unchanged; this is a different time-dependent quote, not a verified change to direct-provider list prices." }, { "slug": "glm-5-3", "name": "GLM-5.3", "catalogId": "z-ai/glm-5.3", "previousDate": "2026-10-08", "currentDate": "2026-10-09", "previous": { "input": 0.049, "output": 3.39, "cached": 0.048 }, "current": { "input": 0.039, "output": 4.8, "cached": 0.038 }, "scheduled": false, "previousScheduleRemoved": false, "note": "These are routed catalog observations, not a verified change to direct-provider list prices. Compare both input and output: one may fall while the other rises." } ], "lifecycle": [ { "model": "Claude Sonnet 4.5", "providerId": "claude-sonnet-4-5-20250929", "status": "Deprecated; retirement scheduled", "announced": "2026-09-30", "retirement": "2026-11-30", "replacement": "claude-sonnet-5-5", "checked": "2026-10-09", "sourceUrl": "https://platform.claude.com/docs/en/about-claude/model-deprecations", "note": "Claude API retirement notice. Partner-operated Bedrock and Google Cloud schedules can differ. Historical benchmarks remain historical; scores are not transferred to the recommended replacement." }, { "model": "GLM-4.5", "providerId": "z-ai/glm-4.5", "status": "Catalog expiration scheduled", "announced": null, "retirement": "2026-12-31", "replacement": null, "checked": "2026-10-09", "sourceType": "catalog", "sourceUrl": "https://openrouter.ai/z-ai/glm-4.5", "note": "The exact OpenRouter entry now declares this expiration date. This is endpoint-specific catalog evidence; it does not establish a direct-provider shutdown or retirement of every hosted copy. No exact cutoff time or replacement is verified." }, { "model": "GLM-4.7", "providerId": "z-ai/glm-4.7", "status": "Catalog expiration scheduled", "announced": null, "retirement": "2026-12-31", "replacement": null, "checked": "2026-10-09", "sourceType": "catalog", "sourceUrl": "https://openrouter.ai/z-ai/glm-4.7", "note": "The exact OpenRouter entry now declares this expiration date. This is endpoint-specific catalog evidence; it does not establish a direct-provider shutdown or retirement of every hosted copy. No exact cutoff time or replacement is verified." }, { "model": "Dots 3 Note Preview free route", "providerId": "dots-studio/dots-3-note-preview:free", "status": "Catalog expiration scheduled", "announced": null, "retirement": "2026-12-31", "replacement": null, "checked": "2026-10-09", "sourceType": "catalog", "sourceUrl": "https://openrouter.ai/dots-studio/dots-3-note-preview:free", "note": "The exact OpenRouter entry now declares this expiration date. This is endpoint-specific catalog evidence; it does not establish a direct-provider shutdown or retirement of every hosted copy. No exact cutoff time or replacement is verified." }, { "model": "GPT-5.3 Codex", "providerId": "gpt-5.3-codex", "status": "Deprecated; retirement scheduled", "announced": "2026-10-01", "retirement": "2027-04-01", "replacement": "gpt-6-sol", "checked": "2026-10-09", "sourceUrl": "https://developers.openai.com/api/docs/deprecations", "note": "Official OpenAI API notice gives six months of notice. Historical results remain attached to the tested model; no successor scores are transferred." }, { "model": "GPT-5.4 nano", "providerId": "gpt-5.4-nano", "status": "Deprecated; retirement scheduled", "announced": "2026-10-01", "retirement": "2027-04-01", "replacement": "gpt-6-luna", "checked": "2026-10-09", "sourceUrl": "https://developers.openai.com/api/docs/deprecations", "note": "Official OpenAI API notice gives six months of notice. Historical results remain attached to the tested model; no successor scores are transferred." }, { "model": "GPT-5.1", "providerId": "gpt-5.1", "status": "Deprecated; retirement scheduled", "announced": "2026-10-01", "retirement": "2027-04-01", "replacement": "gpt-6-sol", "checked": "2026-10-09", "sourceUrl": "https://developers.openai.com/api/docs/deprecations", "note": "Official OpenAI API notice gives six months of notice. Historical results remain attached to the tested model; no successor scores are transferred." }, { "model": "GLM5.2 hosted by Mistral", "providerId": "zai-glm-5-2", "status": "Hosted endpoint deprecated; retirement scheduled", "announced": "2026-09-29", "retirement": "2026-10-31", "replacement": "zai-glm-5-3", "checked": "2026-10-09", "sourceUrl": "https://docs.mistral.ai/resources/changelogs", "note": "This notice applies to the Mistral-hosted endpoint. It does not retire every Z.ai GLM5.2 deployment. The provider states the replacement uses the same price; no benchmark score is transferred." }, { "model": "Antigravity Agent May 2026 preview", "providerId": "antigravity-preview-05-2026", "status": "Managed agent \u00b7 earliest shutdown date reached", "announced": "2026-09-17", "retirement": "2026-10-05", "replacement": "antigravity-preview-09-2026", "checked": "2026-10-09", "sourceUrl": "https://ai.google.dev/gemini-api/docs/deprecations", "note": "Google lists October 5 as the earliest shutdown date for this managed agent, rather than a base model. We have not tested whether the endpoint is already unavailable. Local-tool integrations need the new parameter and file-edit contracts described in the September 17 release notes; changing only the agent ID may be insufficient." }, { "model": "OpenAI Chat Latest", "providerId": "chat-latest", "status": "Mutable alias \u00b7 snapshot updated", "announced": "2026-10-07", "retirement": null, "replacement": null, "checked": "2026-10-09", "sourceUrl": "https://developers.openai.com/api/docs/changelog", "note": "OpenAI updated this chat snapshot on October 7. It remains a moving alias, not a newly named base model or a pinned benchmark version. The provider recommends the GPT-6 family for production APIs; no predecessor score is transferred." }, { "model": "OpenRouter Haiku Latest", "providerId": "~anthropic/claude-haiku-latest", "status": "Routed alias \u00b7 now points to Haiku 5.5", "announced": null, "retirement": null, "replacement": "anthropic/claude-haiku-5.5", "checked": "2026-10-09", "sourceType": "catalog", "sourceUrl": "https://openrouter.ai/api/v1/models", "note": "The exact catalog alias changed from Haiku 4.5 to Haiku 5.5 between the October 7 and October 8 observations. Thinking defaults, tokenization and long-context prices change with it. Pin and test a specific model before depending on this alias; native Anthropic IDs are separate contracts." }, { "model": "Deep Research December 2025 agent", "providerId": "deep-research-pro-preview-12-2025", "status": "Deprecated; shutdown scheduled", "announced": "2026-10-08", "retirement": "2026-10-23", "replacement": "deep-research-preview-04-2026 or deep-research-max-preview-04-2026", "checked": "2026-10-09", "sourceUrl": "https://ai.google.dev/gemini-api/docs/changelog", "note": "This applies to the managed Deep Research agent, not the underlying Gemini model family. Update the agent parameter in interactions.create and test speed versus comprehensiveness on your research tasks." }, { "model": "DeepSeek V4.1 Flash Fast Gateway offer", "providerId": "deepseek/deepseek-v4.1-flash-fast", "status": "Absent from current Gateway catalog", "announced": null, "retirement": null, "replacement": null, "checked": "2026-10-09", "sourceUrl": "https://ai-gateway.vercel.sh/v1/models", "note": "The route was present October 8 and is absent October 9; its Gateway model page now returns 404. This does not establish native model retirement or prove that every account request fails. The historical profile and offer are preserved; verify a currently supported route before using it." } ], "corrections": [ { "checked": "2026-10-04", "note": "OpenRouter corrected the previous free-service FAQ and clarified that context is the serving model\u2019s limit. The routing page preserves unknown fixed prices and exact system identity.", "sourceUrl": "https://openrouter.ai/nvidia/switchyard", "modelId": "nvidia/switchyard" }, { "modelId": "mistral-large-4-high", "checked": "2026-10-08", "sourceUrl": "https://livebench.ai/table_2026_06_25.csv", "note": "LiveBench revised six task values in the existing Large 4 high row. Today\u2019s category scores use the corrected row; the October 7 download preserves its original values. This is a benchmark correction, not a new model release or effort sweep." } ], "serviceModes": { "checked": "2026-10-09", "announced": "2026-10-08", "sourceUrls": [ "https://developers.openai.com/api/docs/changelog", "https://developers.openai.com/api/docs/pricing", "https://developers.openai.com/api/docs/guides/ultrafast-mode" ], "models": [ { "id": "gpt-6.1-sol", "name": "GPT-6.1 Sol", "context": 1050000, "maxOutput": 128000, "modes": [ { "id": "standard", "name": "Standard", "input": 2, "output": 10, "cached": 0.1, "write": 2.5, "longInput": 4, "longOutput": 15.0, "longCached": 0.2, "longWrite": 5.0, "regions": [ "global", "us", "eu" ] }, { "id": "priority", "name": "Fast", "input": 4, "output": 20, "cached": 0.2, "write": 5.0, "longInput": 8, "longOutput": 30.0, "longCached": 0.4, "longWrite": 10.0, "regions": [ "global", "us", "eu" ] }, { "id": "ultrafast", "name": "Ultrafast", "input": 12, "output": 60, "cached": 0.6, "write": 15.0, "longInput": 24, "longOutput": 90.0, "longCached": 1.2, "longWrite": 30.0, "regions": [ "global", "us", "eu" ] } ] }, { "id": "gpt-6-astra", "name": "GPT-6 Astra", "context": 1050000, "maxOutput": 128000, "modes": [ { "id": "standard", "name": "Standard", "input": 10, "output": 50, "cached": 1, "write": 12.5, "longInput": 20, "longOutput": 75.0, "longCached": 2, "longWrite": 25.0, "regions": [ "global", "us", "eu" ] }, { "id": "priority", "name": "Fast", "input": 20, "output": 100, "cached": 2, "write": 25.0, "longInput": 40, "longOutput": 150.0, "longCached": 4, "longWrite": 50.0, "regions": [ "global", "us", "eu" ] }, { "id": "ultrafast", "name": "Ultrafast", "input": 60, "output": 300, "cached": 6, "write": 75.0, "longInput": 120, "longOutput": 450.0, "longCached": 12, "longWrite": 150.0, "regions": [ "global", "us" ] } ] } ] }, "scope": "Reviewed text models, serving offers, composite systems and specialized decision and research models discovered since the September 23 snapshot. These are distinct kinds of systems, not a count of independent base models. Discovery and catalog addition dates are not verified launch dates. Upgrade cost scenarios use selected OpenRouter offers; other documented prices are shown separately." }