Fireworks AI Serverless AI Token Pricing & API Rates
Verified RatesTier 3 DatacenterReal-time verified rates (Updated 2026)
Fireworks AI Serverless AI Token Pricing (5 Models)
| Model | Category | Cold Start | Context | Input ($/1M or Unit) | Output / 1M Tokens | Speed | Actions |
|---|---|---|---|---|---|---|---|
| FLUX.1-dev | IMAGE | 0s Instant | — | $0.025 / img | — | Stateless API | API Docs |
| DeepSeek-V3 | LLM | 0s Instant | 128K | $0.20 | $0.80 | 110 tps | API Docs |
| DeepSeek-R1 | REASONING | 0s Instant | 128K | $0.55 | $2.19 | 95 tps | API Docs |
| Llama-3.3-70B-Instruct | LLM | 0s Instant | 128K | $0.90 | $0.90 | 140 tps | API Docs |
| Qwen-2.5-72B-Instruct | LLM | 0s Instant | 128K | $0.90 | $0.90 | 135 tps | API Docs |
Simulate monthly input and output token volume on Fireworks AI and compare your savings against OpenAI GPT-4o and Claude 3.5.
Open Fireworks AI Cost CalculatorCompare alternative inference platforms like Groq, Together AI, and DeepInfra on token latency, context windows, and pricing.
View Alternatives ComparisonFireworks AI Pricing FAQ & Cost Guide
How does serverless token billing work on Fireworks AI?
Fireworks AI meters exact input (prompt) and output (completion) tokens. You only pay for what your application generates. There is zero baseline server provisioning cost or ongoing hourly commitment.
Does Fireworks AI charge for stopped instances or idle storage?
No. Fireworks AI is a purely serverless platform. When your application receives zero traffic, your hourly billing is strictly $0.00. There are no persistent disk fees or stopped instance retention charges.
Are there hidden egress or bandwidth fees?
Egress on Fireworks AI is billed at Free / Included in Token Rate.
What payment methods does Fireworks AI accept?
Fireworks AI accepts standard credit card billing and enterprise invoicing with verified business accounts.