Agent Mag
SUBSCRIBE
MiMo-V2-Pro
All models
XiaomiXiaomiMiMoReleased 2026-03-18

MiMo-V2-Pro

1.0M context$1.00/M input$3.00/M output1T

MiMo-V2-Pro is Xiaomi's flagship foundation model, featuring over 1 trillion parameters and a context length of 1 million tokens. It is optimized for agentic scenarios and integrates seamlessly with general agent frameworks like OpenClaw. The model is designed to orchestrate complex workflows, drive production engineering tasks, and deliver reliable results, ranking among the global top tier in benchmarks such as PinchBench and ClawBench.

What is MiMo-V2-Pro?

MiMo-V2-Pro is an AI model from Xiaomi that Agent Mag tracks for pricing, context window, modalities, benchmarks, and API compatibility. Builders can use this page to compare MiMo-V2-Pro against other models for agent workflows and production deployments.

Model ID

MiMo-V2-Pro is Xiaomi's flagship foundation model, featuring over 1 trillion parameters and a context length of 1 million tokens. It is optimized for agentic scenarios and integrates seamlessly with general agent frameworks like OpenClaw. The model is designed to orchestrate complex workflows, drive production engineering tasks, and deliver reliable results, ranking among the global top tier in benchmarks such as PinchBench and ClawBench.

Architecture & Specifications
Parameters
1T
Tokenizer
Other
Released
2026-03-18
Modalities
Input
text
Output
text
Supported Parameters
frequency_penaltyinclude_reasoningmax_tokenspresence_penaltyreasoningresponse_formatstoptemperaturetool_choicetoolstop_p
Strengths
  • Extremely large context length of 1 million tokens
  • Optimized for agentic scenarios and workflows
  • Ranks globally in top tier for benchmarks like PinchBench and ClawBench
  • Highly adaptable to general agent frameworks like OpenClaw
  • Reliable performance in complex production engineering tasks
Limitations
  • High hallucination rate (70.1%) in knowledge-based tasks
  • Low performance in research-level physics reasoning (0.3%)
  • Limited accuracy in omniscience tasks (26.8%)
  • Potentially high inference cost due to large parameter count
  • Requires user IDs for anonymity, limiting privacy options
Recommended Use Cases
Integration with agent frameworks like OpenClaw
Orchestrating complex workflows
Driving production engineering tasks
Coding and terminal use in agentic scenarios
Scientific reasoning and graduate-level tasks

More from Xiaomi

MiMo-V2.5-Pro
1.0M ctx$1.00/M

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....

MiMo-V2.5
1.0M ctx$0.400/M

MiMo-V2.5 is a native omnimodal model by Xiaomi. It delivers Pro-level agentic performance at roughly half the inference cost, while surpassing MiMo-V2-Omni in multimodal perception across image and video understanding...

MiMo-V2-Omni
262K ctx$0.400/M

MiMo-V2-Omni is a frontier omni-modal model that natively processes image, video, and audio inputs within a unified architecture. It combines strong multimodal perception with agentic capability - visual grounding, multi-step...

MiMo-V2-Flash
262K ctx$0.090/M

MiMo-V2-Flash is an open-source foundation language model developed by Xiaomi. It is a Mixture-of-Experts model with 309B total parameters and 15B active parameters, adopting hybrid attention architecture. MiMo-V2-Flash supports a...

Related content

Browse the AI models directory

Compare pricing, local installs, context windows, and modality filters across the full model catalog.

Explore AI agent tools

Find frameworks, SDKs, and infrastructure tools that pair with this model in production workflows.

Read Agent Mag model and agent articles

See Agent Mag coverage of model benchmarks, agent frameworks, and deployment patterns.

Data enriched Apr 24, 2026. Pricing from OpenRouter API.