Xiaomi releases MiMo-V2-Pro with 1M context window and 1T+ parameters
Xiaomi released MiMo-V2-Pro on March 18, 2026, a flagship foundation model with over 1 trillion total parameters and a 1,048,576 token context window. The model is priced at $1 per million input tokens and $3 per million output tokens, positioning it as an agent-focused system comparable to top-tier models.
MiMo-V2-Pro at a glance
How MiMo-V2-Pro ranks now
| Benchmark | Score | Rank | Source |
|---|---|---|---|
| Arena Elo | 1448 Elo | 56 of 141 | LMArena, Nov 2026 ↗ |
| GPQA | 87.0% | 70 of 163 | Artificial Analysis, Jul 2026 ↗ |
| SWE-bench Verified | 78.0% | 22 of 80 | the-decoder.com ↗ |
Rank is against every model in our database with a sourced score on that benchmark, each counted once with its best score.
Xiaomi Releases MiMo-V2-Pro With 1M Context Window and 1T+ Parameters
Xiaomi released MiMo-V2-Pro on March 18, 2026, a foundation model featuring over 1 trillion total parameters and a 1,048,576 token context window. The model is priced at $1 per million input tokens and $3 per million output tokens.
Technical Specifications
MiMo-V2-Pro is positioned as Xiaomi's flagship foundation model, designed primarily for agentic scenarios and complex workflow orchestration. The model features:
- Context window: 1,048,576 tokens (1M)
- Parameter count: Over 1 trillion total parameters
- Input pricing: $1 per million tokens
- Output pricing: $3 per million tokens
- Release date: March 18, 2026
Benchmarks and Performance Claims
According to Xiaomi, MiMo-V2-Pro "ranks among the global top tier in the standard PinchBench and ClawBench benchmarks, with perceived performance approaching that of Opus 4.6." The company does not publish specific benchmark scores, instead relying on comparative claims against existing models.
OpenRouter's usage data shows the model handling 319 billion prompt tokens, 1.03 billion completion tokens, and 476 million reasoning tokens in recent tracking periods.
Agentic Focus and Integration
MiMo-V2-Pro is explicitly optimized for agent frameworks, with Xiaomi citing OpenClaw compatibility as a key feature. The company describes the model as "designed to serve as the brain of agent systems, orchestrating complex workflows, driving production engineering tasks, and delivering results reliably."
The model supports reasoning-enabled inference through OpenRouter, allowing access to step-by-step thinking processes via the reasoning parameter in API requests.
Availability and Distribution
MiMo-V2-Pro is accessible through OpenRouter, which handles provider routing and fallback mechanisms to maximize uptime. OpenRouter normalizes API requests and responses across multiple provider implementations.
What This Means
Xiaomi enters the large foundation model market with competitive context window sizing (matching or exceeding most current offerings) and aggressive pricing on the output tier ($3/M output tokens). The focus on agent-optimized design suggests Xiaomi is targeting enterprise automation workflows rather than general-purpose chat applications. The 1M context window places MiMo-V2-Pro in the extended-context category used by models handling document analysis and multi-turn agent reasoning, though specific benchmark data remains unavailable for direct performance comparison. Xiaomi's entry signals continued fragmentation in the foundation model market, with regional players (Chinese tech companies like Xiaomi, Alibaba/Qwen, Baidu, Tencent) establishing independent model lines alongside US-based competitors.
Related Articles
Google launches Gemini 4 Hydrogen: 1M-token context, $1.50 input and $7.50 output per 1M tokens
Google announced Gemini 4 Hydrogen on November 12, 2026, with native 1 million-token input and up to 1 million output tokens. It costs $1.50 per million input tokens and $7.50 per million output tokens, and Google claims it scores 51 on the Artificial Analysis Intelligence Index at its default high thinking level.
Runway releases Praxis-1 robot model with open weights under Apache 2.0, tops World Action Models on RoboLab-120
Runway has released Praxis-1, a robot control model, with open weights and source code under the Apache 2.0 license. It scores 46.2 percent on Nvidia's RoboLab-120 leaderboard, ahead of FLUX 3 Action at 42.9 percent, in tests Runway ran itself.
Meituan releases LongCat-2.5: 1.6T-parameter multimodal MoE with 48B active per token
Meituan's LongCat team has published LongCat-2.5, a native multimodal mixture-of-experts model with 1.6 trillion total parameters and approximately 48 billion activated per token. It targets long-horizon agentic tasks such as codebase development and full-project builds, and ships with W4A4 expert quantization for domestic AI accelerators.
Tencent Hunyuan releases Hy Realtime voice model with tool calling and singing, live in Yuanbao
Tencent Hunyuan announced Hy Realtime on November 13, a real-time voice model that decides on its own when to run a web search and can sing songs on request. It is live in Tencent Yuanbao voice calls and Tencent's customer service, with a Tencent Cloud API to follow. API pricing has not yet been disclosed.
Comments
Loading...