MiMo-V2.5-Pro

3.95
Official Site API / Console
MiMo-V2.5-Pro is a flagship MoE model specialized in long-term agent tasks and complex coding. It stably executes over 1,000 tool calls, reduces token consumption by 40–60% vs. competitors, and is open-sourced under MIT license.
Advertisement 728 × 90
CompanyXiaomi
Context1M
Released2026-04
Updated2026-07-20

MiMo-V2.5-Pro Overview

MiMo-V2.5-Pro is Xiaomi’s flagship model, delivering strong performance in general agentic capabilities, complex software engineering, and long-horizon tasks, with top rankings on benchmarks such as ClawEval, GDPVal, and SWE-bench Pro. It can independently and autonomously complete professional tasks that would take human experts days or weeks, involving more than a thousand tool calls. Its context length of up to 1M makes it well suited for integration with a wide range of agent frameworks.

MiMo-V2.5-Pro Pricing

PlanPriceDescription
Xiaomi API Access Input $0.435 / Output $0.87 per 1M Tokens Supports up to a 1M-token context window and up to 128K maximum output tokens. Built on an approximately 1T-parameter MoE architecture with around 42B active parameters. Cached input pricing is approximately $0.0036 per 1M Tokens.
UltraSpeed Mode Input $1.305 / Output $2.61 per 1M Tokens A high-speed inference mode designed for latency-sensitive applications. Offers faster response performance at approximately 3× the standard API price.
OpenRouter API Access Potentially lower than standard API pricing Access MiMo-V2.5 Pro through OpenRouter and other third-party providers. Some providers may offer promotional pricing or discounts. The model is released under the MIT license.

MiMo-V2.5-Pro Key Features

1. Superior Agentic Capabilities: Optimized for complex agent tasks, scoring 72.9 on τ³-bench, on par with GPT-5.4
2. Excellent Code Generation: Scores 73.7 on MiMo Coding Bench; developed a full compiler in 4.3 hours
3. Efficient MoE Architecture: 1.02T total params, only 42B activated per inference
4. Million-Token Context Window: 1M context for long documents and complex multi-turn tasks
5. Open-Source under MIT: Fully open-source, free for commercial deployment and fine-tuning
6. Ultra-Fast Inference: UltraSpeed version achieves 1000+ tokens/second
7. Token Efficiency: Saves 40%-60% token consumption compared to competitors

Summary

1.AI Agent Developers: Those building complex multi-step automation workflows and agentic applications
2.Software Engineers: Developers needing code generation, code review, and project building assistance
3.Enterprise Users: Organizations requiring long-document processing and complex task automation
4.Open-Source Community & Researchers: Individuals and teams leveraging high-performance open models for research and development
5.AI Product Teams: Teams needing rapid prototyping and high-performance inference

Comments (0)

Leave a comment

AWritten & Reviewed by Last verified 2026-07-20
Advertisement 728 × 90

Xiaomi Model Comparison

Model Context Pricing API Released Global Heat
MiMo-V2.5-Pro
1M Paid YES 2026-04
79/100
MiMo-V2.5
1.1M — YES 2026-04
52/100

Similar Models

MiMo-V2.5
52
Xiaomi MiMo’s next-generation multimodal AI model, combining text, image, video, and audio understanding with advanced reasoning and Agent capabilities.
Xiaomi
Claude Opus 5
100
Anthropic’s flagship Claude model built for complex reasoning, AI Agents, software development, and enterprise knowledge workflows.
Anthropic
GPT-5.5 Pro
99
I’ve been using GPT-5.5 Pro with a few colleagues for the past several weeks, mostly on the kinds of jobs where regular chatbots tend to fall apart: long documents, messy research questions, code debugging, and tasks that need more than one or two steps of reasoning.
OpenAI
GPT-5.5
99
OpenAI flagship model, a multimodal AI natively supporting text, images, and audio.
OpenAI
GPT-5.6 Luna
99
An efficient GPT-5.6 series model optimized for speed, cost efficiency, and intelligent performance across large-scale AI applications.
OpenAI
Claude Opus 4.8
98
Claude Opus 4.8 is Anthropic's most powerful Opus-series model, featuring multimodal input, reasoning, and a 1M-token context window, excelling in complex reasoning and coding.
Anthropic
GPT-5.6 Luna Pro
97
OpenAI’s next-generation efficient AI model, balancing speed, cost, and intelligence for chat, coding, and automation tasks.
OpenAI
Gemini 3.6 Flash
96
Google

Related News

China’s AI App Market Nears 500 Million Users, With Doubao on Top and Qwen Surging

China’s AI App Market Nears 500 Million Users, With Doubao on Top and Qwen Surging

2026-07-15
Read
AI Chip Startup SambaNova Raises $1 Billion at $11 Billion Valuation, Five Months After Previous Funding Round

AI Chip Startup SambaNova Raises $1 Billion at $11 Billion Valuation, Five Months After Previous Funding Round

2026-07-08
Read
Figma Rolls Out Parallel AI Editing, Letting Designers Generate Multiple Concepts Without Breaking Their Flow

Figma Rolls Out Parallel AI Editing, Letting Designers Generate Multiple Concepts Without Breaking Their Flow

2026-07-08
Read