[guillaumegay13]

Automated sync between the Manifest gateway and this catalog (modelparams-agent).

How entries are verified: the local gateway loads this branch's catalog; for every declared param and test value the agent configures it through the gateway's model-params API, sends a request through the proxy pinned to the model, and checks the gateway applied the value upstream and got a 200 from that model. Params that fail are removed from the entry before it is filed.

This PR is rebuilt from main on every run and merged automatically once CI is green.

Added (3)

Entry	Params verified	Dropped	Corrected	Cloned from
mistral/mistral-large-4.yaml	max_tokens temperature top_p random_seed presence_penalty frequency_penalty safe_prompt	stop response_format.type	temperature min→0.1	mistral/mistral-large-2512.yaml
mistral/mistral-large-4-0.yaml	max_tokens temperature top_p random_seed presence_penalty frequency_penalty safe_prompt	stop response_format.type	temperature min→0.1	mistral/mistral-large-2512.yaml
opencode-go/space-bunny-subscription.yaml	max_tokens temperature top_p top_k presence_penalty frequency_penalty response_format.type	—	—	opencode-go/space-bunny-free-subscription.yaml

Evidence

models/mistral/mistral-large-4.yaml

• baseline: 200 from mistral/mistral-large-4 (api_key)
• max_tokens ✅ 1
• temperature: declared min 0 refused (HTTP 400: top_p must be 1 when using greedy sampling.); 0.1 accepted → min corrected to 0.1
• temperature ✅ 0.1 (min corrected from 0), 1.5
• top_p ✅ 1, 0.01
• random_seed ✅ 0
• presence_penalty ✅ 0, -2, 2
• frequency_penalty ✅ 0, -2, 2
• safe_prompt ✅ true, false
• ❌ dropped stop: no testable value (string without default)
• ❌ dropped response_format.type: response_format.type="json_object" → response is not JSON

models/mistral/mistral-large-4-0.yaml

• baseline: 200 from mistral/mistral-large-4-0 (api_key)
• max_tokens ✅ 1
• temperature: declared min 0 refused (HTTP 400: top_p must be 1 when using greedy sampling.); 0.1 accepted → min corrected to 0.1
• temperature ✅ 0.1 (min corrected from 0), 1.5
• top_p ✅ 1, 0.01
• random_seed ✅ 0
• presence_penalty ✅ 0, -2, 2
• frequency_penalty ✅ 0, -2, 2
• safe_prompt ✅ true, false
• ❌ dropped stop: no testable value (string without default)
• ❌ dropped response_format.type: response_format.type="json_object" → response is not JSON

models/opencode-go/space-bunny-subscription.yaml

• baseline: 200 from opencode-go/opencode-go/space-bunny (subscription)
• max_tokens ✅ 1
• temperature ✅ 0, 1.9
• top_p ✅ 0.01, 1
• top_k ✅ 1
• presence_penalty ✅ -2, 2
• frequency_penalty ✅ -2, 2
• response_format.type ✅ "text", "json_object"

Needs a human: params now refused (2)

Re-verification found these declared params refused. Not removed — the param guard blocks removals and dropping a param is a maintainer's call.

• models/mistral/codestral-latest.yaml response_format.type: response_format.type="json_schema" → HTTP 400: Schema response format type requires a json_schema
• models/mistral/mistral-medium-3.yaml response_format.type: response_format.type="json_schema" → HTTP 400: Schema response format type requires a json_schema

Marked retired (2)

No longer listed by the provider's live /models for N consecutive daily runs.

• models/fireworks/glm-5p2.yaml (missing n/a — not served runs)
• models/openai/gpt-3.5-turbo-1106.yaml (missing n/a — not served runs)

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
modelparams.dev	Ready	Preview	Oct 7, 2026 5:42am UTC