StepFun's sparse Mixture of Experts model that activates 11B of 196B parameters per token, providing a balance between capability and efficiency.
- Strengths
- Its MoE architecture enables high capability while maintaining low inference costs through selective parameter activation.
- Best for
- Workloads requiring strong reasoning and language understanding where latency and throughput efficiency matter.
- Limitations
- As a sparse MoE model, it may show less consistent performance on tasks that don't align well with its expert specializations compared to dense models.
Input / 1M
$0.1
Output / 1M
$0.3
Cached input / 1M
—
Context window
262K
Price history
Input (solid)Output (dashed)
Price change
- 30d
- in increased 11.1% out decreased 0.0%
- 90d
- in increased 11.1% out decreased 0.0%
- 1y
- in increased 11.1% out decreased 0.0%
- Since launch
- in increased 11.1% out decreased 0.0%
Snapshots
| Effective | Input | Output | Cached in | Note | Source |
|---|---|---|---|---|---|
| 30 Jun 2026 | $0.1 | $0.3 | — | Imported from OpenRouter | openrouter.ai |
| 11 Jun 2026 | $0.09 | $0.3 | $0.02 | Imported from OpenRouter | openrouter.ai |