Is AIgateway a drop-in replacement for Fireworks AI?

Yes if you're using Fireworks AI's OpenAI-compatible surface. Change the base URL to https://api.aigateway.sh/v1 and swap the API key — every other line of code stays the same. Streaming format is unchanged.

Will my existing model slugs work?

Some slugs match exactly, others differ. Check /models for the AIgateway equivalent of each model you call. Fireworks AI's accounts/fireworks/models/llama-v4-scout-instruct → AIgateway uses provider/model.

How long does migration take?

60 seconds if you run `aig migrate fireworks` (manually, since CLI automation isn't shipped for this source yet). Test suite should be green immediately because the OpenAI surface is preserved.

What's different about AIgateway's pricing?

5% per call (our entire revenue model) + payment-processor passthrough at top-up (Stripe's standard rate). No monthly fee, no seat fee, no per-model surcharge. Cached requests get a 50% discount. $5 free credits on signup, redeemable on a curated edge-tier shortlist (Kimi K2.6, FLUX-2 Klein, Whisper Turbo, Aura 2, Gemma 4 vision, BGE-M3, Llama Guard 3), valid for 7 days. Topups never expire and unlock the full catalog.

View as /docs.md

Migrate

Migrate from Fireworks AI

Q: Can I get a credit match for my Fireworks AI invoice?

Yes — email your last Fireworks AI invoice to switch@aigateway.sh and we match the credit (up to $1,000) within 24 hours. See /switch/fireworks for details.

Drop-in for Fireworks' OpenAI-compatible API. Bigger catalog (frontier closed models + every modality), same edge latency, lower platform fee on top-ups.

CLI automation for Fireworks AI isn't shipped yet — you'll do this manually below. PRs welcome at packages/aig-cli/src/commands/migrate.ts.

TL;DR — three lines

# 1. set your AIgateway key
export AIGATEWAY_API_KEY="sk-aig-..."

# 2. change the base URL
base_url="https://api.aigateway.sh/v1"  # was https://api.fireworks.ai/inference/v1

# 3. search-and-replace the env var name

What changes, line-by-line

The table below is what aig migrate fireworks rewrites. If you're doing it by hand, do these in order.

Thing	Fireworks AI	AIgateway
Base URL	https://api.fireworks.ai/inference/v1	https://api.aigateway.sh/v1
API key env var	FIREWORKS_API_KEY	AIGATEWAY_API_KEY
Model slug	accounts/fireworks/models/llama-v4-scout-instruct	meta/llama-4-scout (see /models)
Catalog	Open-weight + a few hosted closed models	900+ across every modality
Per-deployment configs	Custom serverless deployments	BYO routing rules in dashboard

Before / after — Python (OpenAI SDK)

BEFORE · Fireworks AI

from openai import OpenAI

client = OpenAI(
    base_url="https://api.fireworks.ai/inference/v1",
    api_key=os.environ["FIREWORKS_API_KEY"],
)
r = client.chat.completions.create(
    model="anthropic/claude-opus-4.7",
    messages=[{"role": "user", "content": "Hi"}],
)

AFTER · AIgateway

from openai import OpenAI

client = OpenAI(
    base_url="https://api.aigateway.sh/v1",
    api_key=os.environ["AIGATEWAY_API_KEY"],
)
r = client.chat.completions.create(
    model="anthropic/claude-opus-4.7",
    messages=[{"role": "user", "content": "Hi"}],
)

Behavioral notes

Model slugs differ — Fireworks uses accounts/<org>/models/<name>; AIgateway uses provider/model. Map via /models.
Frontier models (GPT-5.4, Claude Opus 4.7, Gemini 3.1 Pro) are available in AIgateway but not Fireworks.
Custom serverless model deployments: not supported in AIgateway today; reach out via /enterprise if you need this.

Credit match

Email your last invoice from Fireworks AI to switch@aigateway.sh and we match the credit up to $1,000 (24-hour turnaround). Or read the full pitch at /switch/fireworks.