[codex] add Fireworks provider support

[chai1any]

What changed

• added Fireworks AI as a first-class API-key provider across auth, runtime resolution, model catalogs, setup, doctor, status, and auxiliary client routing
• added Fireworks model metadata and models.dev mapping so provider detection and context-window fallbacks behave consistently
• documented Fireworks usage, environment variables, provider routing, and fallback-provider support in the README and Docusaurus docs
• extended provider, setup, runtime, and models.dev tests to cover Fireworks aliases and curated models

Why

Hermes already supports several OpenAI-compatible providers. Fireworks uses the same integration shape, but it was missing from provider registration, setup flows, runtime/provider detection, and user-facing docs.

Impact

• users can configure FIREWORKS_API_KEY or FIREWORKS_BASE_URL and run Hermes with --provider fireworks
• interactive setup now offers Fireworks alongside the existing provider choices
• provider selection, auxiliary-provider defaults, and fallback docs stay aligned with the existing terminology and UX

Validation

• python -m pytest tests/ -q --ignore=tests/integration --tb=short -n auto
• python -m pytest tests/test_api_key_providers.py tests/test_runtime_provider_resolution.py tests/hermes_cli/test_model_validation.py tests/test_setup_model_selection.py tests/agent/test_models_dev.py -q --ignore=tests/integration --tb=short -n auto
• cd website && npm ci
• cd website && source ../.venv/bin/activate && npm run lint:diagrams && npm run build

[rickross]

does this also add the Fire Pass version of Kimi-K2.5 to the list of models that support vision? That seems to be something separate from just adding the provider.

[rickross]

Wow, well done! I hope they'll merge this one pronto! Thank you.

[spa5k]

Hi, any progress on this?

[alt-glitch]

Likely duplicate of #6054 and #15332 — same scope: add Fireworks AI as built-in provider.

[alt-glitch]

Likely duplicate of #6054 and #15332 — same scope: add Fireworks AI as built-in provider.

[teknium1]

Fireworks AI support landed via PR #62593. This contribution was part of the Fireworks provider cluster credited in the salvage PR; thank you for the implementation and review work.