Inference pricing
Rates are USD per million tokens. Input, output and cache-read tokens have separate rates. Paid fleet inference requires a funded account. Rates and model availability can change; custom provider arrangements may differ.
Public pricing source · Read as Markdown
Retrieved at . This is the retrieval time, not the source's last-updated time.
| Model | Input | Output | Cache read |
|---|---|---|---|
| EigenLabs/Qwen3.8-27B-4bit-mtp | $0.035 | $2.2 | $0.0175 |
| gemma-4-26b | $0.042 | $0.22 | $0.021 |
| gemma-4-26b-8bit | $0.042 | $0.22 | $0.021 |
| gemma-4-26b-qat-4bit | $0.042 | $0.22 | $0.021 |
| gpt-oss-20b | $0.018 | $0.09 | $0.009 |
| mimo-v2.6-flash-mopd | $0.1 | $0.28 | $0.05 |
| nvidia-nemotron-3.5-lightning | $0.025 | $0.12 | $0.0125 |
| qwen3.5-35b-a3b | $0.065 | $0.75 | $0.0325 |
| Qwen3.5-9B | $0.08 | $0.13 | $0.04 |
| qwen3.6-35b-a3b-vl-mtp-mxfp8 | $0.05 | $0.7 | $0.025 |
| qwen3.8-flash-next | $0.13 | $0.4 | $0.065 |
| qwen3-vl-30b-a3b-instruct | $0.09 | $0.4 | $0.045 |
| ternary-bonsai-2-27b | $0.075 | $0.5 | $0.0375 |
Fallback rates
Where the API applies a fallback rate: input $0.05, output $0.2, and cache read $0.025 per million tokens.
Using the API
Model-specific pricing does not guarantee a model is available to serve a request. Use the authenticated model list to discover models, and check your account balance in the Console before sending paid inference requests.