AI Agent StoreInception · Native API verified
Mercury Decide
Route a support request, choose a category or score urgency. Send shared information and named questions; the service returns probabilities and predefined answers.
First reviewed 2026-10-20 · Checked 2026-11-03. Mentioned in the October 19 Mercury 3 announcement; an exact separate launch day is not verified.
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
Token prices measure the small pieces of text the model reads or produces. Capacity is the amount the service accepts in a request; check the limits below for any fallback to another model. A dash means the rate is unverified.
Exact provider API ID: mercury-decide. Questions are evaluated independently against shared information. Choice questions allow 1–255 options and scores 1–10 levels. Images require base64 data, not remote URLs. Up to 16 images, 10 MiB each, must fit within the 20 MiB total JSON request. Extraction can return matching text and its character positions for 1–26 fields per question, with confidence and an indication that a field was not stated. It selects from the input rather than writing new text. Requests using extraction cannot include images; use a separate request for image decisions.
Use /v1/decisions with mercury-decide. Chat completions, streaming and tool calling are unsupported. Validate probability thresholds on your own examples.
A context-token limit and maximum question count are not confirmed. The 20 MiB body allowance is not a token limit. Image encoding adds roughly one third to file size.
Output: Choices, yes/no probabilities, ordered scores, or exact text spans with character positions. Compare decision models, limits and input costs ↗
The launch input price is $0.02 per million billed tokens, half the regular $0.04. Output is not charged. The model table lists cached input at zero; cache eligibility and a promotion end date are not verified. Include retries when comparing the total cost.
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. Check whether confidence scores match the answers you can verify.
Compare caching and long-context costs ↗