MiMo-V2.6-Pro
Output 125 tokens/sFirst token 1.3 sContext 1M tokens
Top 0 cheapest providers (Adjusted $/task)
Within the active global provider, residency and confidentiality filters.
No per-token pricing matches the active global filters.
Composite
1 of 7 inputs6 radar axes: DesignArena's two boards share one94.8
AA Coding —Coding Agent v1.4 —AA Intelligence 46.3AA Agentic —Epoch ECI —Software ECI —DesignArena —/—
Radar: percentile among all models measured on each input; a gap means not measured.
Benchmark sheet
17 of 230 registered benchmark versions · bars show the percentile among all models measured on each benchmark.
17 of 18 values are Xiaomi's own claims (†), not independent measurements; a matching independent result replaces a claim as soon as one exists.
Agentic
Agents' Last Examdeveloper's claim31.6%† self-reported by the developer
Agents' Last Exam published 2026-09-22 → · Published board — Agents' Last Exam result reported by Xiaomi for MiMo-V2.6-Pro in the 2026-09-22 launch benchmark table; the source does not supply a complete independent reproduction recipe.
Automation Bench v1.0.6developer's claim53.1%† self-reported by the developer
Automation Bench v1.0.6 v1.0.6 → · Published board — Automation Bench v1.0.6 result reported by Xiaomi for MiMo-V2.6-Pro in the 2026-09-22 launch benchmark table; the source does not supply a complete independent reproduction recipe.
GDPVal 2.1 (AA) v2.1developer's claim1,673 Elo† self-reported by the developer
GDPVal 2.1 (AA) v2.1 → · Published board — GDPVal 2.1 (AA) result reported by Xiaomi for MiMo-V2.6-Pro in the 2026-09-22 launch benchmark table; the source does not supply a complete independent reproduction recipe.
JobBenchdeveloper's claim62.0%† self-reported by the developer
JobBench published 2026-09-22 → · Published board — JobBench result reported by Xiaomi for MiMo-V2.6-Pro in the 2026-09-22 launch benchmark table; the source does not supply a complete independent reproduction recipe.
OSWorld-Verifieddeveloper's claim82.0%† self-reported by the developer
OSWorld-Verified published 2026-09-22 → · Published board — OSWorld-Verified result reported by Xiaomi for MiMo-V2.6-Pro in the 2026-09-22 launch benchmark table; the source does not supply a complete independent reproduction recipe.
Terminal Bench 2.1 v2.1developer's claim89.9%† self-reported by the developer
Terminal Bench 2.1 v2.1 → · Published board — Terminal Bench 2.1 result reported by Xiaomi for MiMo-V2.6-Pro in the 2026-09-22 launch benchmark table; the source does not supply a complete independent reproduction recipe.
Terminal Bench 4.0 v4.0developer's claim34.9%† self-reported by the developer
Terminal Bench 4.0 v4.0 → · Published board — Terminal Bench 4.0 result reported by Xiaomi for MiMo-V2.6-Pro in the 2026-09-22 launch benchmark table; the source does not supply a complete independent reproduction recipe.
Toolathlon-verifieddeveloper's claim76.9%† self-reported by the developer
Toolathlon-verified published 2026-09-22 → · Published board — Toolathlon-verified result reported by Xiaomi for MiMo-V2.6-Pro in the 2026-09-22 launch benchmark table; the source does not supply a complete independent reproduction recipe.
Coding
CyberGymdeveloper's claim94.0%† self-reported by the developer
CyberGym published 2026-09-22 → · Published board — CyberGym result reported by Xiaomi for MiMo-V2.6-Pro in the 2026-09-22 launch benchmark table; the source does not supply a complete independent reproduction recipe.
DeepSWE v1.1developer's claim71.9%† self-reported by the developer
DeepSWE v1.1 v1.1 → · Published board — DeepSWE v1.1 result reported by Xiaomi for MiMo-V2.6-Pro in the 2026-09-22 launch benchmark table; the source does not supply a complete independent reproduction recipe.
ExploitBenchdeveloper's claim47.9%† self-reported by the developer
ExploitBench published 2026-09-22 → · Published board — ExploitBench result reported by Xiaomi for MiMo-V2.6-Pro in the 2026-09-22 launch benchmark table; the source does not supply a complete independent reproduction recipe.
ExploitGymdeveloper's claim17.8%† self-reported by the developer
ExploitGym published 2026-09-22 → · Published board — ExploitGym result reported by Xiaomi for MiMo-V2.6-Pro in the 2026-09-22 launch benchmark table; the source does not supply a complete independent reproduction recipe.
MiMo Code Benchdeveloper's claim63.2%† self-reported by the developer
MiMo Code Bench published 2026-09-22 → · Published board — MiMo Code Bench result reported by Xiaomi for MiMo-V2.6-Pro in the 2026-09-22 launch benchmark table; the source does not supply a complete independent reproduction recipe.
MiMo Cyber Benchdeveloper's claim81.7%† self-reported by the developer
MiMo Cyber Bench published 2026-09-22 → · Published board — MiMo Cyber Bench result reported by Xiaomi for MiMo-V2.6-Pro in the 2026-09-22 launch benchmark table (81.7; Xiaomi's model card and technical report print 80.2 — the dated launch-table value is shown); the source does not supply a complete independent reproduction recipe.
ProgramBenchdeveloper's claim26.5%† self-reported by the developer
ProgramBench published 2026-09-22 → · Published board — ProgramBench result reported by Xiaomi for MiMo-V2.6-Pro in the 2026-09-22 launch benchmark table; the source does not supply a complete independent reproduction recipe.
SEC Bench Prodeveloper's claim66.3%† self-reported by the developer
SEC Bench Pro published 2026-09-22 → · Published board — SEC Bench Pro result reported by Xiaomi for MiMo-V2.6-Pro in the 2026-09-22 launch benchmark table; the source does not supply a complete independent reproduction recipe.
Reasoning
AA Intelligence Index9746.3
AA Intelligence Index published 2026-09-21 → · Published board — Artificial Analysis publishes this board without a version number; we keep the date each result was retained.
Vision
MiMo Visual Codingdeveloper's claim72.3%† self-reported by the developer
MiMo Visual Coding published 2026-09-22 → · Published board — MiMo Visual Coding result reported by Xiaomi for MiMo-V2.6-Pro in the 2026-09-22 launch benchmark table; the source does not supply a complete independent reproduction recipe.
Missing coverage · 217 benchmark versions
No result does not mean a zero, or that the model was never tested. Collection failures and disputed versions retain their distinct status.
Token offers by platform · 0 offers (Adjusted $/task)
Click any underlined price to see how it is estimated and where each input comes from. How we calculate adjusted cost.
0 offers within the active global filters; “—” means the catalog is active but no public token price is available.
No token offers match the active global filters.