MiMo-V2.6-Flash is a text language model from Xiaomi built on a Mixture-of-Experts architecture with 309B total parameters and 15B activated per token, supporting a 1M token context window.
- Strengths
- The sparse MoE design reduces computational overhead while the hybrid attention mechanism handles long-range dependencies efficiently across the extended context window.
- Best for
- Processing long-form text and maintaining context over extended sequences where a smaller per-token parameter footprint is acceptable.
- Limitations
- As a text-only model, it does not handle images, video, or multimodal inputs; for those use cases or higher throughput demands, MiMo-V2.5-Pro or MiMo-V2.6-Pro-UltraSpeed may be more suitable.
Input / 1M
$0.14
Output / 1M
$0.28
Cached input / 1M
$0.0028
Context window
1.05M
Price history
Snapshots
| Effective | Input | Output | Cached in | Note | Source |
|---|---|---|---|---|---|
| 22 Sep 2026 | $0.14 | $0.28 | $0.0028 | Imported from OpenRouter | openrouter.ai |