Xiaomi: MiMo-V2.6-Flash

Audio input Video input Text input Image input Text output
Author's Description

MiMo-V2.6-Flash is an open-source foundation model developed by Xiaomi. Built on a Mixture-of-Experts architecture with 309B total parameters and 15B activated per token, it employs a hybrid attention mechanism for...

Key Specifications
Cost
$$$
Context
1M
Parameters
309B (Rumoured)
Released
Sep 21, 2026
Speed
★★
Ability
★★★★
Reliability
★★★
Supported Parameters

This model supports the following parameters:

Top P Structured Outputs Temperature Tool Choice Reasoning Include Reasoning Presence Penalty Response Format Stop Tools Frequency Penalty Max Tokens
Features

This model supports the following features:

Tools Structured Outputs Response Format Reasoning
Performance Summary

Xiaomi's MiMo-V2.6-Flash, an open-source foundation model with a Mixture-of-Experts architecture, demonstrates competitive response times, ranking in the 53rd percentile across benchmarks. It offers cost-effective solutions, placing in the 68th percentile for pricing. Notably, the model exhibits exceptional reliability with a 99% success rate, indicating consistent and stable performance. In terms of specific benchmarks, MiMo-V2.6-Flash achieves perfect accuracy in Hallucinations (Baseline), effectively acknowledging uncertainty, and stands out as the most accurate model at its price point and speed for this category. It also performs strongly in General Knowledge (99.5% accuracy) and Email Classification (99.0% accuracy), where it ranks in the 90th percentile for the latter. Its Reasoning capabilities are solid at 92.0% accuracy. While its Instruction Following (69.0% accuracy) and Ethics (97.0% accuracy, 29th percentile) scores are more moderate, its Coding performance (91.0% accuracy) is respectable, though its duration for this task is in the lower 28th percentile. Mathematics also shows a 92.0% accuracy, but with a longer duration. Key strengths include its high reliability, excellent hallucination prevention, and strong performance in general knowledge and email classification. A potential area for improvement could be optimizing speed for computationally intensive tasks like Coding and Mathematics.

Model Pricing

Current Pricing

Feature Price (per 1M tokens)
Prompt $0.14
Completion $0.28
Input Cache Read $0.0028

Price History

Available Endpoints
Provider Endpoint Name Context Length Pricing (Input) Pricing (Output)
Xiaomi
Xiaomi | xiaomi/mimo-v2.6-flash-20260921 1M $0.14 / 1M tokens $0.28 / 1M tokens
DeepInfra
DeepInfra | xiaomi/mimo-v2.6-flash-20260921 1M $0.14 / 1M tokens $0.28 / 1M tokens
Relace
Relace | xiaomi/mimo-v2.6-flash-20260921 1M $0.14 / 1M tokens $0.28 / 1M tokens
Venice
Venice | xiaomi/mimo-v2.6-flash-20260921 1M $0.175 / 1M tokens $0.35 / 1M tokens
Novita
Novita | xiaomi/mimo-v2.6-flash-20260921 1M $0.14 / 1M tokens $0.28 / 1M tokens
GMICloud
GMICloud | xiaomi/mimo-v2.6-flash-20260921 1M $0.14 / 1M tokens $0.28 / 1M tokens
Darkbloom
Darkbloom | xiaomi/mimo-v2.6-flash-20260921 1M $0.1 / 1M tokens $0.28 / 1M tokens
Io Net
Io Net | xiaomi/mimo-v2.6-flash-20260921 1M $0.159 / 1M tokens $0.3 / 1M tokens
InferenceNet
InferenceNet | xiaomi/mimo-v2.6-flash-20260921 1M $0.14 / 1M tokens $0.28 / 1M tokens
InferenceNet
InferenceNet | xiaomi/mimo-v2.6-flash-20260921 1M $0.14 / 1M tokens $0.28 / 1M tokens
Benchmark Results
Benchmark Category Reasoning Strategy Free Executions Accuracy Cost Duration
Other Models by xiaomi