fix(chat): handle dynamic output limits

[actus7]

Resumo

• usa maxOutputTokens do modelo quando disponível
• infere limites por família/provider e usa fallback seguro quando desconhecido
• preserva limites retornados por APIs OpenAI-compatible de modelos
• injeta max_tokens dinâmico no chat e em /v1 após Smart Routing
• detecta finish_reason=length e mostra Continuar gerando só quando houve truncamento real

Testes

• pnpm test -- lib/model-output-limits.test.ts lib/chat-stream.test.ts server/tests/v1-auto-lane-scoring.test.ts server/tests/v1-auto-routing.test.ts server/lib/routing/complexity-scorer.test.ts lib/chat-utils.test.ts
• pnpm typecheck
• pnpm lint (passa com warning existente em .remember/tmp/last-ndc.ts)

[gemini-code-assist]

Caution

The consumer version of Gemini Code Assist on GitHub has been sunset. All code review activity has officially ceased.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[vercel]

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

Project	Deployment	Actions	Updated (UTC)
modelhub	Ready	Preview, Comment	Jul 29, 2026 3:30pm