We are at re:Invent 2026Book my merch

Fireworks AI

Every Fireworks token, accounted for.

Cloud Intelligence™ imports your Fireworks AI cost and usage every six hours: serverless inference, dedicated deployments and fine-tuning, by model, GPU, user and API key. Right next to the rest of your AI and cloud spend.

Admin API key · 180 days of history · Refreshed every 6 hours

Fireworks AI · October
$84,320
Serverless inference58%Dedicated deployments31%Fine-tuning11%
Tokens
9.6B
GPU-hours
1,840
Cached input
34%
By model and accelerator
GPT OSS 120B serverless$24,180
NVIDIA_B300_288GB dedicated$19,760
Qwen3 235B serverless$14,930
DeepSeek V3 serverless$9,840
Llama fine-tune training$9,280

Trusted by teams serving production inference on Fireworks

Akamai
Blumira
Chess.com
Current
OneFootball
Promptly
Recurly
Lusha
Mux

what we import

Fireworks data, in the dimensions you already use.

Every charge maps to standard Cloud Analytics dimensions, plus Fireworks labels for model, deployment type, usage type, cached tokens, accelerator, user and API key.

Cloud AnalyticsFireworks AIExample
ProviderFireworks AIFireworks AI
Billing accountAccount IDmy-account
ServiceProduct lineServerless Inference
SKUModel or acceleratorGPT OSS 120B
Plus Fireworks AI labels
ModelDeployment typeUsage typeCachedAcceleratorUserAPI key
Refreshed every 6 hours · 180 days of history on connect

what you can do with it

From the Fireworks bill to a decision

Cost by key and user

Slice by model, deployment type, usage type, user or API key.

Cached vs uncached

See how much input is served from cache, and what it saves.

Serverless vs dedicated

Token spend and GPU-hours on dedicated deployments, side by side.

GenAI Intelligence

Fireworks next to your other AI providers in one dashboard.

Budgets and alerts

Guardrails on inference spend before the invoice lands.

Allocations

Put Fireworks spend on teams and products, by key or user.

live in hours

Connect Fireworks AI in three steps

1

Create an admin API key.

In the Fireworks dashboard. Admin keys unlock cost per user and per key.

2

Connect in the catalog.

Enter your account ID and key, then test the connection.

3

Reports in hours.

Up to 180 days of history imported. We email you when it's ready.

Read-only billing access. Disconnect any time.

Start free

Frequently asked
questions

How does Cloud Intelligence™ connect to Fireworks AI?

With your Fireworks account ID and an API key from an account administrator. Enter both on the Fireworks AI card in the integrations catalog, test the connection and save.

Why does the key need to belong to an administrator?

Cost per user and per API key comes from a Fireworks endpoint that requires administrator access. With a non-admin key, we still import your overall billing data, but skip the per-user and per-key breakdown.

How fresh is the data?

We refresh Fireworks AI data every six hours. Fireworks reports costs daily, and a day's data is typically final about one day later. On first connection we import up to 180 days of history.

Which dimensions and labels do I get?

Provider, billing account, service (serverless inference, dedicated deployments, fine-tuning) and SKU (model, or accelerator for dedicated deployments), plus labels for model, deployment type, usage type, cached tokens, accelerator and account. Usage is in tokens for serverless and fine-tuning, and GPU-hours for dedicated deployments.

Are credits and discounts included?

No. We import gross rated usage. Credits, discounts, taxes and invoice adjustments are not exposed by the Fireworks AI billing API.

Can I see Fireworks next to OpenAI, Anthropic and my cloud bills?

Yes. Fireworks AI costs appear in the GenAI Intelligence dashboard alongside your other AI providers, and in the same reports, budgets and allocations as your cloud spend.

Connect Fireworks AI. Know what every token costs.

Free to start. Read-only. Reports in hours.