---
id: "xiaomi/mimo-v2.6-pro"
also: ["xiaomi/mimo-v2.6-pro-20260921", "XiaomiMiMo/MiMo-V2.6-Pro-RL", "hf/XiaomiMiMo/MiMo-V2.6-Pro-RL", "requesty/xiaomi/mimo-v2.6-pro", "vercel/xiaomi/mimo-v2.6-pro"]
card_usd_per_m: "0.435 in, 0.87 out, at xiaomi/fp8"
cheapest_list_usd_per_m: "0.3262 in, 0.6525 out, at gmicloud/bf16 on vercel"
cheapest_effective_usd_per_m: "0.0282 in, 0.6521 out, at gmicloud/bf16 on openrouter"
context: 1050000
max_output: 131072
input: ["text", "image", "video", "audio"]
output: ["text"]
released: "2026-09-21"
api: ["/api/v1/chat/completions", "/api/v1/completions", "/api/v1/messages", "/api/v1/responses"]
params: ["frequency_penalty", "include_reasoning", "logit_bias", "logprobs", "max_tokens", "min_p", "presence_penalty", "reasoning", "repetition_penalty", "response_format", "seed", "stop", "structured_outputs", "temperature", "tool_choice", "tools", "top_k", "top_logprobs", "top_p"]
endpoints: 15
gateways: "vercel (6), openrouter (5), requesty (3), hf (1)"
no_training: 9
trains: 4
retains_prompts: 10
intelligence: 29.9000
---
# Xiaomi: MiMo-V2.6-Pro

MiMo-V2.6-Pro is the flagship of Xiaomi's MiMo-V2.6 series: a sparse Mixture-of-Experts model with 1.02T total and 42B activated parameters, natively omnimodal across text, image, video and audio, with a 1M-token context window. Trained with large-scale mixed reinforcement learning and groupwise agentic grading, it is built for demanding agentic work including long-horizon coding, multi-tool agents and cybersecurity tasks.