AI Agent StoreLiquid AI · Download verified
Liquid d1 2B
Route a support ticket, read a value from a document or locate objects in an image using a model you run yourself.
Released 2026-10-27 · First reviewed 2026-10-28 · Checked 2026-11-03.
We still need comparable independent tests. You can compare the confirmed features and limits below, but we cannot yet give this version an Agent Fit or Practical Value rating. Test it on records with known answers, including cases that should go to a person.
What you can use today
You download the model files and supply the hardware and software to run them. Capacity is the amount of information a request can contain; settings and available memory can limit what fits.
Exact download ID: LiquidAI/d1-2B. Returns yes/no probabilities, choices, scores, extracted text, numbers, dates and image boxes. It can select more than one answer and estimate a number with a range. Extended answer formats require Transformers; llama.cpp supports only yes/no, choice and score. Finding boxes requires exactly one picture.
Downloadable under the LFM Open License v1.0, which limits commercial use by revenue. Review the full license for your organization. Use the model’s matching software and preprocessing; quality and device-speed claims on its card are publisher tests. Test confidence thresholds on examples from your own work before allowing automatic actions.
The card describes 128K context; the configuration sets 128,000 positions. The shared information, media and questions must fit that capacity; this model does not generate a reply. Questions are processed in groups of 16 by default; this is not a maximum number of questions per call. The publisher warns that answers become less reliable beyond about 64K tokens of shared information.
Output: Probabilities, choices, scores, extracted values and image boxes; no generated reply. Compare decision models, limits and input costs ↗
Running this download needs hardware, memory and maintenance. A model-specific hosted price and measured local operating cost are not established here.
Provider release and specifications ↗
Check availability and plans ↗
More setup and licensing details ↗
Test it before switching
Give each model the same examples, tools and definition of success. Record correct outcomes, response time, retries and total cost. Include the time and hardware needed to run it.
Compare caching and long-context costs ↗