feat: add groq/qwen3-32b parameters

[guillaumegay13]

Adds parameter coverage for groq/qwen3-32b (Groq-served Qwen3-32B). Addresses #84.

New provider groq (api_key), plus a lobe-icons logo mapping.

Parameters (10)

max_completion_tokens, temperature, top_p, frequency_penalty, presence_penalty, seed, stop, reasoning_effort, reasoning_format, response_format.type.

Verified against the live Groq API

Every parameter was smoke-tested against POST /openai/v1/chat/completions with model: qwen/qwen3-32b (Groq's API model string; the catalog id flattens the qwen/ prefix, consistent with how nvidia/* is stored). Each returned 200; an unsupported param like top_k returns 400, so acceptance is real.

• reasoning_effort values are none / default — confirmed by the API's own validation error: "reasoning_effort must be one of none or default". low/medium/high are rejected on qwen3 (those are for gpt-oss).
• frequency_penalty / presence_penalty are accepted (200), so they're listed even though Groq notes they're currently behavioral no-ops.

Validated locally: npm run validate, npm test, npm run guard:params, npm run typecheck, npm run build all pass.

[vercel]

@guillaumegay13 is attempting to deploy a commit to the Manifest Team on Vercel.

A member of the Team first needs to authorize it.

[vercel]

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

Project	Deployment	Actions	Updated (UTC)
modelparams.dev	Ready	Preview	Jul 14, 2026 1:47pm