Providers · Beta

Give each Fireworks AI workload its own limits.

Connect one Fireworks account. Give each app an approved serverless chat model, an immediate spending limit, rate ceilings, and a compatible fallback.

Create a free workspace Try the no-card Test Lab

How the connection works

How Fireworks AI works with the gateway

Check eligible models without inference

The connection test reads the Fireworks catalog for serverless-eligible models with the encrypted API key. It does not send a prompt or generate tokens merely to prove the key can reach the account.

Stop each workload at its limit

Every app receives a scoped gateway key, dollar ceiling, request pace, token pace, and concurrency cap. Fireworks' account-wide request envelope and adaptive per-model token limits remain the upstream ceiling.

Keep the serverless boundary explicit

The adapter accepts Chat Completions and rejects unsupported protocol boundaries. A route must use the full ID of a serverless-eligible model; a dedicated deployment is not silently treated as a shared serverless endpoint.

Current shared controls

What every supported connection includes

Encrypted BYOK credentials that remain hidden after setup Stable model aliases and compatible route targets Organization and key budgets, rate limits, and concurrency limits Observe, Shadow, and Enforce modes with metadata-only evidence

Test the exact Fireworks model ID, Chat Completions payload, streaming behavior, returned usage, and adaptive upstream limits before production approval.

Client boundary

AIGHQ_BASE_URL=https://api.aigatewayhq.com/v1
AIGHQ_API_KEY=from-your-workspace
AIGHQ_MODEL=company-approved-fast

Each Fireworks-backed application uses the gateway URL, its scoped workload key, and a stable model alias. Administrators can change an eligible serverless chat model or Fireworks API key centrally without distributing the provider credential to every app.

Implemented boundary

Supported protocols

Select Fireworks AI and enter an API key issued to a service account or user. The connection check reads only serverless-eligible models without running inference. Then map the full Fireworks model ID to a stable alias.

POST /v1/chat/completions OpenAI-compatible streaming GET /v1/accounts/fireworks/models connection test

Compatibility follows the provider's current published interface. Review the provider documentation before approving Fireworks AI for production.