PRICING / USD
Pay for tokens.
Keep the math simple.
No platform subscription for shared inference. Input and output are metered separately, in millions of tokens.
Shared inference
| Model | Input / 1M | Output / 1M |
|---|---|---|
| Stride 8B | $0.12 | $0.20 |
| Summit 32B | $0.65 | $0.90 |
| Forge 14B Code | $0.28 | $0.42 |
| Embed 768 | $0.02 | — |
| Embed 1536 | $0.05 | — |
½
Offline workload? Use batches.
Chat and code batches receive 50% off interactive rates. Plan for a 24-hour processing window.
MONTHLY ESTIMATE
Size up your usage.
Estimated token costUSD / month · excluding taxes
Request access ↗Reserve your own lane.
For sustained traffic, custom concurrency and region requirements. Capacity is quoted by model, replica count and commitment.
Discuss reserved capacity →