Skip to main content
This page is the contract-level map of which providers are used in runtime paths.

LLM Paths

WorkflowPrimaryFallbackKeys
Chat generation (/api/chat*)Groq chat_model (default llama-3.3-70b-versatile)OpenAI chat_model_openai (default gpt-4o-mini)GROQ_API_KEY, OPENAI_API_KEY
Policy-style direct responsesOpenAI chat_model_openaiGroq chat_modelOPENAI_API_KEY, GROQ_API_KEY
Learnings extractionAnthropic learnings_model (default claude-sonnet-4-5-20250929)Groq chat_modelANTHROPIC_API_KEY, GROQ_API_KEY
Project extractionGroq chat_modelnoneGROQ_API_KEY
Project idea generationAnthropic (chat_model_anthropic or project override) / OpenAI (mode-dependent)Groq fallback pathANTHROPIC_API_KEY, OPENAI_API_KEY, GROQ_API_KEY
Transcript cleanupOpenAI gpt-4ononeOPENAI_API_KEY
Snippet verificationOpenAI chat_model_openainoneOPENAI_API_KEY

Retrieval and Search Providers

CapabilityProvider / ModelKeys
EmbeddingsOpenAI embedding_model (default text-embedding-3-small)OPENAI_API_KEY
RerankingCohere rerank-v4.0-proCOHERE_API_KEY
Explicit web searchTavilyTAVILY_API_KEY

Ingestion Providers

CapabilityProviderKeys
YouTube transcript + metadataSupadataSUPADATA_API_KEY
PDF OCR extractionMistral OCR (mistral_ocr_model)MISTRAL_API_KEY
Dropbox media transcription (optional flow)Gladia + Dropbox APIsGLADIA_API_KEY, DROPBOX_ACCESS_TOKEN

Runtime Behavior Notes

  • Chat uses circuit-breaker-aware fallback (groq -> openai) for resilience.
  • Missing Cohere key degrades reranking to fallback ordering (no hard outage).
  • Deterministic branches (clear_meta, some ambiguity paths) can complete without LLM generation.

Change Policy

If a provider/model choice changes:
  1. update runtime code/config
  2. update this page in the same PR
  3. run tests + eval checks
  4. update related architecture/contracts docs if behavior changed