MiMo-V2.6-Pro

MiMo-V2.6-Pro is Xiaomi's flagship entry in the MiMo-V2.6 series, positioned as an omni-modal reasoning model intended for complex projects, long-horizon tasks, high-stakes professional work, cybersecurity, and research. Released on September 22, 2026 alongside two siblings — mimo-v2.6-flash for high-frequency professional calls and mimo-v2.6-pro-ultraspeed for latency-sensitive workloads — it represents Xiaomi's push beyond consumer electronics into competitive foundation model territory. The model is distributed as open source under the permissive MIT license, making it accessible for both commercial deployment and research use, and its design emphasizes handling inputs across images and text within the same system.

Within the MiMo family, MiMo-V2.6-Pro sits at the top of the V2.6 lineup as the highest-capacity reasoning option, contrasted with the lighter, lower-cost flash variant and the speed-optimized ultraspeed variant. Its intended fit is demanding workflows where sustained reasoning across multimodal information matters more than raw call volume, including security-sensitive and research-oriented applications. Practitioners choosing it should expect a model tailored to depth and rigor over throughput, with open-weight availability that supports self-hosting, auditing, and customization in environments where licensing terms and model provenance are deciding factors.

Vultrmimo-v2.6-pro-rlmimo

Open provider page Provider docs

Quick Info

Provider: Vultr
Model key: mimo-v2.6-pro-rl
Release date: Sep 22, 2026
Last updated: Sep 22, 2026
Input modalities
Output modalities
Capabilities

Cost

Input token cost: $0.40
Output token cost: $0.80

Limits

Output tokens: 131,072 tokens
Context window: 1,048,576 tokens

Transparent token rates

Compare MiMo-V2.6-Pro pricing

Rates are shown per one million tokens. Combined means one million input plus one million output tokens.

Browse this family

Latest news about MiMo-V2.6-Pro

Vultr

Coverage

MiMo-V2.6-Pro from Xiaomi has debuted as the top-performing open-weight model on Artificial Analysis' Intelligence Index with a score of 46, tied with Grok 4.7 and ahead of Grok 4.6 (44) and Gemini 3.8 Flash (41), and beating DeepSeek V4.1 Flash (39) and V4.1 Pro (36). The model is open-weight and MIT-licensed, with Xiaomi pricing its API at $0.435 per million input tokens and $0.87 per million output tokens, and Artificial Analysis measuring about $0.13 per Intelligence Index task. The flagship supports a 1-million-token context window with native text, image, audio, and video input, and runs at roughly 134 tokens per second output speed. It is joined by the smaller MiMo-V2.6-Flash, priced at $0.14/$0.28 per million input/output tokens, which retains the same 1M context and multimodal capabilities while targeting high-volume production workloads. Developers can download weights from Hugging Face for free customization and self-hosting.

Vultr

CoverageBenchmark

BenchLM's independent aggregation places MiMo-V2.6-Pro at a capability score of 74.2/100, ranking 9th out of 210 tracked models, with its strongest performance in coding (rank 9 of 143, 94th percentile) and agentic tasks (rank 16 of 117, 87th percentile). The model is priced at $0.43 input and $0.87 output per million tokens, with a $0.004 cached-input rate and a blended rate of about $0.65. Reported throughput is approximately 41 tokens per second with a first-token latency of 53.23 seconds, and the model supports a 1-million-token context window. Benchmark coverage is partial across the 14 verified rows, with reasoning, multimodal, knowledge, multilingual, instruction-following, and math categories not yet measured, leaving room for future ranking shifts as more evidence is published.