Xiaomi Releases MiMo-V2.6-Flash: Open-Source MoE Model with 1M-Token Context, $0.14/$0.28 per 1M Tokens
Xiaomi has released MiMo-V2.6-Flash, an open-source Mixture-of-Experts model with 309B total parameters and 15B activated per token, featuring a 1M-token context window and native multimodal capabilities. Priced at $0.14 per 1M input tokens and $0.28 per 1M output tokens, it targets agentic coding and long-horizon task workflows.
MiMo-V2.6-Flash — Quick Specs
Xiaomi has released MiMo-V2.6-Flash, an open-source foundation model built on a Mixture-of-Experts (MoE) architecture with 309 billion total parameters, of which 15 billion are activated per token. The model is available now via OpenRouter and other providers, priced at $0.14 per 1 million input tokens and $0.28 per 1 million output tokens.
Specifications
MiMo-V2.6-Flash features a 1-million-token context window and native multimodal capabilities, handling text alongside visual inputs within a unified architecture. According to Xiaomi, the model employs a hybrid attention mechanism designed to improve computational efficiency relative to standard transformer attention at this parameter scale.
The company positions the model for agentic workflows, claiming strong performance across coding, visual, general, and research scenarios. Xiaomi states the model excels at "complex, long-horizon tasks with robust generalization across a diverse range of agent harnesses" — though no specific benchmark scores (e.g., SWE-bench, MMLU) were disclosed for this particular release, unlike its predecessor MiMo-V2-Flash, which Xiaomi says ranked #1 among open-source models on SWE-bench Verified and SWE-bench Multilingual.
On OpenRouter, the model shows a P50 latency of 1.04 seconds and throughput of 116 tokens per second from Xiaomi's own hosting, with 99.81% uptime recorded over the platform's tracking window.
Where it fits in Xiaomi's lineup
MiMo-V2.6-Flash is the latest entry in Xiaomi's expanding MiMo model family, which now spans multiple tiers:
- MiMo-V2.6-Pro: flagship model exceeding 1 trillion parameters, $0.435/$0.87 per 1M tokens
- MiMo-V2.6-Pro-UltraSpeed: same 1T-parameter checkpoint as Pro but claims ~10x output speed, $4.35/$8.70 per 1M tokens
- MiMo-V2.5-Pro: prior flagship, $0.3045/$0.609 per 1M tokens
- MiMo-V2.5: omnimodal model, $0.119/$0.238 per 1M tokens
- MiMo-V2-Flash: predecessor MoE model with the same 309B/15B parameter split but a 256K context window
The V2.6-Flash release keeps the same MoE parameter configuration as V2-Flash while expanding context from 256K to 1M tokens and adding native multimodal input handling.
What this means
Xiaomi is pricing MiMo-V2.6-Flash aggressively — at $0.14/$0.28 per 1M tokens, it undercuts most Western frontier-lab offerings by a wide margin while still claiming performance suitable for agentic coding and multi-step tool use. The 15B active-parameter design keeps inference costs low despite the 309B total parameter count, a pattern now common among Chinese labs (DeepSeek, Alibaba Qwen, Moonshot AI) competing on cost-per-token rather than raw parameter count alone.
No independent benchmark verification accompanies this release, and Xiaomi has not published a technical report with specific scores for MiMo-V2.6-Flash, unlike claims made for the earlier V2-Flash version. Buyers evaluating the model for production agentic workloads should treat the "top-tier" and "robust generalization" language as vendor claims pending third-party evaluation.
Related Articles
OpenBMB's MiniCPM-V 4.7 35B-A3B MoE claims 13+ tokens/s on iPhone 17 Pro Max in 3 GB
OpenBMB has released MiniCPM-V 4.7 35B-A3B, the first mixture-of-experts model in the MiniCPM-V series, with open weights on Hugging Face. The company claims it runs at 13+ tokens/s using 3 GB of memory on an iPhone 17 Pro Max, and that it beats Qwen3.6 35B-A3B on overall score in less inference time in Instruct mode.
StepFun releases Step 5 Preview: 600B MoE with 1M context at $1/$2.70 per 1M tokens
StepFun has listed Step 5 Preview, a sparse Mixture-of-Experts model with 600B total and 27B active parameters and a 1.0M-token context window. It is priced at $1 input and $2.70 output per 1M tokens on OpenRouter. StepFun positions it as its flagship model for agentic work.
Fish Audio releases Drama 3, a 35B-parameter MoE text-to-speech model with 16-speaker support
Fish Audio has released Drama 3, a text-to-speech model with 35B total parameters and 3B active per token, built for multi-character, long-form audio. It supports up to 16 speakers per text, voice cloning, and streaming. Fish Audio reports first audio in under 100 ms on a single H200.
Unbiased releases Pareto 26.9: 262K-context composite model at $2.50/$7.50 per 1M tokens
Unbiased has released Pareto 26.9, a multimodal composite model with a 262K-token context window, priced at $2.50 per 1M input tokens and $7.50 per 1M output tokens. It is listed on OpenRouter with a Oct 13, 2026 release date. A cheaper 1.0M-context preview, Pareto 26.10, is already listed alongside it.
Comments
Loading...