Xiaomi Launches MiMo-V2.5 With 1M Context Window at $0.40 per Million Input Tokens
Xiaomi released MiMo-V2.5 on April 22, 2026, a native omnimodal model with a 1,048,576 token context window. The model is priced at $0.40 per million input tokens and $2 per million output tokens, positioning it as a cost-efficient alternative for agentic applications requiring multimodal perception across image and video understanding.
Xiaomi Launches MiMo-V2.5 With 1M Context Window at $0.40 per Million Input Tokens
Xiaomi released MiMo-V2.5 on April 22, 2026, a native omnimodal model featuring a 1,048,576 token context window priced at $0.40 per million input tokens and $2 per million output tokens.
Specifications and Pricing
MiMo-V2.5 offers:
- Context window: 1,048,576 tokens (1M)
- Input pricing: $0.40 per million tokens
- Output pricing: $2 per million tokens
- Release date: April 22, 2026
According to Xiaomi, the model delivers "Pro-level agentic performance at roughly half the inference cost" compared to unspecified alternatives, though the company has not provided independent benchmark scores to verify these claims.
Technical Capabilities
Xiaomi describes MiMo-V2.5 as a "native omnimodal model" designed for multimodal perception across image and video understanding tasks. The company claims the model surpasses its predecessor, MiMo-V2-Omni, in multimodal perception, though specific benchmark comparisons were not disclosed.
The 1M context window is designed to handle complete documents, extended conversations, and complex task contexts in a single inference pass. Xiaomi positions this capability as particularly suited for integration with agent frameworks.
Availability
The model is currently available through OpenRouter, which routes requests across multiple providers to optimize uptime and handle varying prompt sizes. OpenRouter supports the model's reasoning capabilities through a dedicated reasoning parameter that exposes step-by-step thinking processes via a reasoning_details array in API responses.
What This Means
MiMo-V2.5 enters an increasingly competitive omnimodal model market with a clear value proposition: extended context at lower input pricing than many enterprise-tier alternatives. At $0.40 per million input tokens, it undercuts several comparable models while offering a 1M context window—a specification typically reserved for premium tiers.
The focus on agentic workflows suggests Xiaomi is targeting developers building autonomous systems that require sustained reasoning across multimodal inputs. However, without published benchmark scores on standard evaluation sets like MMLU, VQAv2, or video understanding benchmarks, independent assessment of the model's claimed performance advantages remains difficult. The model's effectiveness will ultimately be determined by real-world deployment results in production agent systems.
Related Articles
Alibaba releases Qwen 3.8, a 2.4 trillion parameter open-weight model claiming second place behind Fable 5
Alibaba has released Qwen 3.8, a 2.4 trillion parameter open-weight model that the company claims trails only Fable 5. The multimodal model processes images, videos, and documents, with a preview available through Alibaba's platforms at 10 percent of standard pricing.
Thinking Machines releases Inkling: 975B-parameter MoE model with Apache 2.0 license, first major US open-weight multimo
Thinking Machines Lab released Inkling, a mixture-of-experts model with 975B total parameters and 41B active parameters, trained on 45 trillion tokens across text, images, audio, and video. The Apache 2.0-licensed model supports up to 1M context and debuts alongside Inkling-Small (276B-A12B), marking what observers call the strongest US-based open-weight release to date.
Meituan launches LongCat 2.0: 1.6T parameter MoE model with 1M+ context window at $0.30 per 1M input tokens
Meituan has released LongCat 2.0, a sparse mixture-of-experts language model with 48 billion active parameters out of 1.6 trillion total. The model features a 1,049,000 token context window and costs $0.30 per 1M input tokens and $1.20 per 1M output tokens.
Alibaba previews Qwen3.8 with 2.4 trillion parameters, claims second place without benchmark data
Alibaba unveiled Qwen3.8 at the World Artificial Intelligence Conference in Shanghai, claiming the 2.4 trillion parameter model ranks second only to Anthropic's Fable 5. The company provided no benchmark scores, model card, or independent verification to support the claim.
Comments
Loading...