MiMo Flash model for multimodal coding agents and long-context automationMiMo-V2.6-Flash by xiaomi is offered by 35 providers on this page. Public prices are shown for 30 of them. Its official list price is $0.14 per 1M input tokens and $0.28 per 1M output tokens. The lowest paid channel is Vultr at $0.10 / $0.25 per 1M, about 1.3× below the list price. It also has 2 free ($0) channels and 3 covered by a paid subscription; free tiers usually carry rate limits, and subscription-covered access bills $0 per token only after the subscription fee.
Artificial Analysis rates it 26.7 on the Intelligence Index, with 45.7 for Strategy & Ops, its strongest domain. Against its lowest blended price of $0.138 per 1M, that is roughly 194 index points per dollar, which is the value ratio the leaderboards rank on. Median output speed is 38 tokens per second, with 3.26s to the first token. Running one task of the Artificial Analysis suite costs about $0.1285, which reflects how many tokens its reasoning consumes rather than the unit price alone.
The context window is 1,048,576 tokens at the reference host, but hosts report different limits, from 200,000 to 1,050,000, so the usable window depends on the provider you pick. It supports reasoning, tool use, and structured output. Accepted input modalities are Text, Image, Audio, and Video. The weights are open, so it can also be self-hosted or served through a gateway of your choice. Providers report the capability flags inconsistently, so verify a specific feature against the host you plan to use.
Quality is independently evaluated by Artificial Analysis. Speed/latency are model-level medians.
| Provider | Tier | Input | Output | Cache read | Cache write | Context | Output limit | Status |
|---|---|---|---|---|---|---|---|---|
| Xiaomi Token Plan (Singapore) | Gateway | Subscription | $0 | — | 1,048,576 | 131,072 | ||
| AMD MiMo-V2.6-Flash | Cloud | Free | $0 | — | 1,048,576 | 131,072 | ||
| Xiaomi Token Plan (China) | Gateway | Subscription | $0 | — | 1,048,576 | 131,072 | ||
| OpenCode mimo-v2.6-flash-free | Gateway | Free | $0 | — | 200,000 ⚠ | 32,000 | ||
| Xiaomi Token Plan (Europe) | Gateway | Subscription | $0 | — | 1,048,576 | 131,072 | ||
| Vultr mimo-v2.6-flash-rl | Cloud | $0.10 | $0.25 | — | — | 1,048,576 | 131,072 | |
| Kilo Gateway | Gateway | $0.10 | $0.28 | $0.0028 | — | 1,048,576 | 131,072 | |
| Pareto Inference | Gateway | $0.13 | $0.25 | $0.0025 | — | 1,048,576 | 131,072 | |
| DevPass (LLM Gateway) | Gateway | $0.14 | $0.27 | $0.0036 | — | 1,048,576 | 131,072 | |
| LLM Gateway | Gateway | $0.14 | $0.27 | $0.0036 | — | 1,048,576 | 131,072 | |
| XiaomiOfficial | First-party | $0.14 | $0.28 | $0.0028 | — | 1,048,576 | 131,072 |
Sorted by blended price (input×0.75 + output×0.25) asc. The official channel always shows regardless of rank. Whether a gateway's low price is actually usable can't be verified.
Interleaved thinking (reasoning between tool calls) is declared by 12 of 35 providers.
5 / 35 providers expose no reasoning control (reasoning_options: []).
5 more channels offer $0 (Xiaomi Token Plan (Singapore), AMD, Xiaomi Token Plan (China) etc.); free tiers usually have rate limits and no SLA, excluded from ranking.
| Name | Conditions | Score | Metric | Source |
|---|---|---|---|---|
| DeepSWE | v1.1 | 67.9 | score | Source ↗ |
| ProgramBench | — | 26 | score | Source ↗ |
| MiMo Code Bench | — | 61.2 | score | Source ↗ |
| AutomationBench | v1.0.6 | 52.3 | score | Source ↗ |
| Toolathlon-Verified | — | 73.6 | score | Source ↗ |
| Agents' Last Exam | — | 27.6 | score | Source ↗ |
| Terminal-Bench | v4.0 | 28.8 | score | Source ↗ |
| Terminal-Bench | v2.1 | 87.6 | score | Source ↗ |
| OSWorld-Verified | — | 80.8 | score | Source ↗ |
| JobBench | — | 61.2 | score | Source ↗ |
| CyberGym | — | 95.1 | score | Source ↗ |
| MiMo Cyber Bench | — | 77.2 | score | Source ↗ |
| ExploitGym | — | 6 | score | Source ↗ |
| ExploitBench | — | 25.3 | score | Source ↗ |
| SEC-Bench Pro | — | 47.5 | score | Source ↗ |
| MiMo VisualCoding | — | 71.5 | score | Source ↗ |
The same benchmark scores very differently across harness / dataset, so the qualifying conditions must be shown together.
Individual evaluations run by Artificial Analysis, on the same reasoning tier as the intelligence score above. Each benchmark has its own task set and scale (percent correct, points or an index), so rows are not comparable with one another.