Fireworks AI Pricing in 2026
4 plans.
Enterprise reserved instance
Ask
Billed Contact sales
Reserved GPUs; multi-region deployments, custom optimizations, better performance, and BYOC compatibility
Managed training
Ask
Billed Per 1M training tokens; rates vary by model size and training method
LoRA SFT, LoRA DPO, Full Param SFT, and Full Param DPO; rates range by model size
On-demand deployments
Ask
Billed Per GPU second, with listed per-minute and per-hour rates
H100 and H200: $0.134/minute or $8.00/hour; B200: $0.217/minute or $13.00/hour; B300: $0.250/minute or $15.00/hour; GB300: $0.334/minute or $20.00/hour
Serverless inference — embeddings
Ask
Billed Usage-based, per input token
Prices vary by embedding model parameter count
Read from Fireworks AI's pricing page on 7 Oct 2026.
What the alternatives cost
| Product | Free | Paid from |
|---|---|---|
| Dify | Free plan | $590/yr |
| Cursor | Free plan | $8/mo |
| Replit | Free plan | $18/mo |
| OpenHands | Free plan | — |
| Flowise | Free plan | $35/mo |
| Langflow | Free plan | — |