Summary
Add Fireworks AI as a built-in OpenAI-compatible provider preset in Zero, using the serverless /inference/v1 chat-completions endpoint with an ordinary API key (FIREWORKS_API_KEY).
Motivation
Fireworks is a common inference provider for open-weight coding and agent models. Zero already ships many OpenAI-compatible presets; Fireworks fits the same pattern and lets users activate it with:
export FIREWORKS_API_KEY=...
zero providers setup fireworks --set-active
Proposed behavior
- Catalog ID:
fireworks
- Endpoint:
https://api.fireworks.ai/inference/v1
- Auth:
FIREWORKS_API_KEY
- Default + curated picker models must be serverless-capable (not on-demand deployment IDs), because the documented API-key flow does not provision deployments
- Suggested default:
accounts/fireworks/models/kimi-k2p7-code
- Map provider →
fireworks-ai for models.dev hydration
Related PR
Notes for maintainers
Please add the issue-approved label if this scope is accepted so the linked community PR can proceed under CONTRIBUTING.md.
Summary
Add Fireworks AI as a built-in OpenAI-compatible provider preset in Zero, using the serverless
/inference/v1chat-completions endpoint with an ordinary API key (FIREWORKS_API_KEY).Motivation
Fireworks is a common inference provider for open-weight coding and agent models. Zero already ships many OpenAI-compatible presets; Fireworks fits the same pattern and lets users activate it with:
export FIREWORKS_API_KEY=... zero providers setup fireworks --set-activeProposed behavior
fireworkshttps://api.fireworks.ai/inference/v1FIREWORKS_API_KEYaccounts/fireworks/models/kimi-k2p7-codefireworks-aifor models.dev hydrationRelated PR
Notes for maintainers
Please add the
issue-approvedlabel if this scope is accepted so the linked community PR can proceed under CONTRIBUTING.md.