stealthmodels.

MODEL FIELD GUIDE

Mimo V2.6 Flash: Beach

Low-cost multimodal Flash service with a million-token context. Xiaomi also publishes MIT-licensed RL and MOPD checkpoints for local deployment.

Drawn by this model01 / 01
Beach benchmark result generated by this model
SVG BENCHMARKBeachView result

A good fit for

Multimodal agents and high-volume API work where token cost matters, with a separate choice of released checkpoint for self-hosting.

Context window
1,000,000tokens[1]
Active parameters
15Bparameters[4] [6]
License
MIT[4] [5]
Model factsArchitecture, context & capabilities
Developer
Xiaomi[1] [2]
Origin
China[3]
Family
MiMo V2.6[1] [2]
Context window
1,000,000 tokens[1]
Maximum output
128,000 tokens[1]
License
MIT[4] [5]
Architecture
Released Flash-RL checkpoint: sparse Mixture-of-Experts; hosted Flash and MOPD variants are distinct artifacts[4] [6]
Total parameters
309B[4] [6]
Active parameters per token
15B[4] [6]
Input
text, image, video, audio[1]
Output
text[1]
API access & pricingOpenRouter · $0.14 input / $0.28 output · per 1M tokens

Xiaomi MiMo API; input 0.14 USD/1M; cached read 0.0028/1M; output 0.28/1M

[1]
  • OpenRouterObserved 2026-10-06
    Input $0.14Output $0.28Cached read $0.0028

    per 1M tokens

    xiaomi/mimo-v2.6-flash[7]
  • Xiaomi MiMo APIObserved 2026-10-06
    Input $0.14Output $0.28Cached read $0.0028

    per 1M tokens

    Available

    mimo-v2.6-flash[1]
Run it locallyWeights, memory & deployment

multi-GPU datacenter

[5] [4]
  • Parameter-based FP8 weight-only lower bound309 GB storage · FP8 equivalentEstimated[4]
Measurements & sources7 primary references · Speed & serving conditions

StealthMark score

StealthMark measures fluid and visual intelligence of AI models.

General and Visual

General Intelligence

Perspective, anatomy, correct shadows, purposeful detail and physical consistency.

Visual Intelligence

Likeness, anatomy, detail, light, materials, composition and artistic expression.

Judging

Frontier Models blindly judge the AI generation unbiased through an intricate process. Our researchers control for mistakes.

Overall score

Overall combines weighted task scores on a scale from 0 to 100. Repeating a task does not increase its weight. Estimated means the total includes an estimate for one missing scene.

Benchmaxxing

Benchmaxx flags models that score unusually well on the familiar Pelican task compared with their other benchmark results. The percentage measures that imbalance, not the probability of training contamination. Flagged Pelican scores are excluded from Overall. Pelican never counts toward General or Visual.

We keep some prompts private to discourage test-specific optimization.

Scoring categories

Prompt & likeness

Subjects, actions and likeness.

Anatomy & construction

Coherent bodies, joints and machinery.

Space & placement

Perspective, scale and contact.

Light & reflections

Lighting, shadows and reflections.

Purposeful detail

Clear, purposeful small features.

Artistry & character

Composition, expression and drawing skill.

Materials & effects

Surfaces, texture and movement.

Visual integrity

Clean shapes, layers and edges.

BENCHMARK RESULTS

Compare results

This link opens the same results in the same order.

StealthMark benchmark results

Reset benchmark preferences?

Return to the default view, clear comparison selections and restore the advertisement.