Gate News message, April 27 — Xiaomi’s MiMo team has open-sourced the MiMo-V2.5 series of large language models under MIT license, supporting commercial deployment, continued training, and fine-tuning. Both models feature a 1 million token context window. MiMo-V2.5-Pro is a pure-text mixture-of-experts (MoE) model with 1.02 trillion total parameters and 42 billion active parameters, while MiMo-V2.5 is a native multimodal model with 310 billion total parameters and 15 billion active parameters, supporting text, image, video, and audio understanding.
MiMo-V2.5-Pro targets complex agent and programming tasks. In ClawEval benchmarks, it achieved 64% Pass@3 while consuming approximately 70,000 tokens per task trajectory—40% to 60% fewer tokens than Claude Opus, Gemini 3.1 Pro, and GPT-5.4. The model scored 78.9 on SWE-bench Verified. In a demonstration, V2.5-Pro independently implemented a complete SysY-to-RISC-V compiler for a Peking University compiler course project in 4.3 hours with 672 tool calls, achieving a perfect score of 233/233 on hidden test sets.
MiMo-V2.5 is designed for multimodal agent scenarios, equipped with a dedicated vision encoder (729 million parameters) and audio encoder (261 million parameters), scoring 62.3 on the Claw-Eval general subset. Both models employ a hybrid architecture combining sliding window attention (SWA) and global attention (GA), paired with a 3-layer multi-token prediction (MTP) module for accelerated inference. Model weights are available on Hugging Face.
Alongside the open-source release, the MiMo team launched the “Orbit Quadrillion Token Creator Incentive Program,” offering 100 quadrillion tokens free over 30 days to global users. Individual developers, teams, and enterprises can apply via the program page with an evaluation cycle of approximately 3 business days; approved benefits are distributed as Token Plans or direct credits, compatible with tools like Claude Code and Cursor.
Disclaimer: The information on this page may come from third parties and does not represent the views or opinions of Gate. The content displayed on this page is for reference only and does not constitute any financial, investment, or legal advice. Gate does not guarantee the accuracy or completeness of the information and shall not be liable for any losses arising from the use of this information. Virtual asset investments carry high risks and are subject to significant price volatility. You may lose all of your invested principal. Please fully understand the relevant risks and make prudent decisions based on your own financial situation and risk tolerance. For details, please refer to
Disclaimer.
Related Articles
DeepSeek Increases Registered Capital by 50%, Founder Liang Wenfeng's Stake Rises to 34%
Gate News message, April 28 — According to Qichacha data, DeepSeek (Hangzhou Deepseek AI Basic Technology Research Co., Ltd.) increased its registered capital from 10 million yuan to 15 million yuan. Founder Liang Wenfeng's registered capital contribution rose from 1 million yuan to 5.1 million yuan
GateNews5m ago
Meta Reserves 1GW+ Solar and Energy Storage for AI Data Centers
Meta has signed energy deals with Overview Energy and Noon Energy to secure power capacity for its artificial intelligence data centers, according to The Economic Times. The company reserved up to 1 gigawatt
CryptoFrontier9m ago
Zhipu Z.ai Extends GLM Coding Plan Triple Quota Promotion to June 30
Gate News message, April 28 — Zhipu Z.ai announced an extension of its GLM Coding Plan triple quota promotion from the original April 30 deadline to June 30, covering both GLM-5.1 and GLM-5-Turbo models.
The promotion is available during Eastern Time 6:00 AM to 2:00 AM the following day. GLM Coding
GateNews24m ago
TSMC Accelerates Expansion at Double Pace, Five 2nm Fabs to Ramp Production in 2026
Gate News message, April 28 — Taiwan Semiconductor Manufacturing Company (TSMC) is accelerating its capacity expansion at double the historical pace to meet surging demand from artificial intelligence and high-performance computing, according to Hou Yongqing, senior vice president at TSMC. The
GateNews29m ago
GitHub Copilot Switches to Token-Based Billing Starting June 1, Monthly Fees Unchanged
Gate News message, April 28 — GitHub announced that all Copilot plans will transition to token-based billing effective June 1, 2026. The existing "premium request" pricing model will be replaced by GitHub AI Credits, with usage calculated based on actual token consumption (including input, output, a
GateNews35m ago
Chinese Regulators Block Meta's $2 Billion Manus Acquisition
Gate News message, April 28 — Chinese government agencies have blocked Meta's $2 billion acquisition of AI agent platform Manus, announced yesterday (April 27). The deal, initially completed in December, faced
GateNews39m ago
