Inference
EN

MiMo-V2.6-Pro vs MiniMax M3.1 Flash

Change models
MiMo-V2.6-ProXiaomi
MiniMax M3.1 FlashMiniMax
IndexAll categories, out of 100.
70.2 ±4.7 best
69.3 ±5.4
GapIndex points below the top.
Top
−0.9 · no clear gap
CoverageIndex weight that has results.
80%
80%
SpeedOutput tokens per second.
31/s
100/s best
EloLMArena rating and rank.
1492
—
ReleasedMonth of release.
2026-09
2026-10
Input / 1MUS dollars per 1M input tokens.
$0.435
$0.15 best
Output / 1MUS dollars per 1M output tokens.
$0.87
$0.60 best
1K chats2K tokens in, 500 out, each.
$1.31
$0.60 best
1K documents20K tokens in, 1K out, each.
$9.57
$3.60 best
1K agent tasks200K tokens in, 20K out, each.
$104.40
$42.00 best
ContextMaximum tokens in one request.
1.05M
1.05M
AgenticMulti-step tasks with tools.
72.1 best
71.8
CodingCode writing and repair.
66.6
69.4 best
ReasoningLogic problems and puzzles.
68.3 best
67.9
MultimodalTasks with images and text.
—
58.1
KnowledgeFacts and expert knowledge.
64.2
68.9 best
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
—
—
MathMath problems.
71.3
—

Benchmarks

26 shared
Only MiMo-V2.6-Pro: 32Only MiniMax M3.1 Flash: 10
MiMo-V2.6-Pro
MiniMax M3.1 Flash
Artificial Analysis Humanity's Last ExamKnowledge
49.4% best14 / 222
48.0%17 / 222
Artificial Analysis Long Context ReasoningReasoning
86.3% best3 / 217
79.7%45 / 217
Critical Physics TasksReasoning
26.6% best19 / 214
22.3%25 / 214
Artificial Analysis Omniscience AccuracyKnowledge
34.8%67 / 164
45.1% best37 / 164
GDPval-AA normalizedAgentic
59.2% best9 / 133
47.9%31 / 133
Vibe Code Bench v1.1Coding
85.2% best16 / 123
73.9%45 / 123
Artificial Analysis SciCodeCoding
60.9% best8 / 114
53.4%39 / 114
Legal Research BenchLegal
47.1% best16 / 90
34.1%50 / 90
Tax Agent BenchFinance
32.4% best22 / 82
32.1%23 / 82
LiveBench Data AnalysisReasoning
72.3% best49 / 80
71.4%52 / 80
LiveBench ReasoningReasoning
79.6%59 / 80
82.9% best48 / 80
Terminal-Bench 4.0Agentic
31.3% best24 / 61
25.3%28 / 61
DeepSWEAgentic
71.9%18 / 59
73.7% best13 / 59
Agent Arena steerabilityAgentic
1.5 best25 / 54
−2.636 / 54
Agent Arena task outcomeAgentic
5.0 best16 / 54
−0.734 / 54
LiveBench Coding 2.0Coding
81.5%30 / 52
83.9% best26 / 52
LiveBench DesignAgentic
78.5% best18 / 52
74.0%31 / 52
LiveBench ResearchAgentic
76.7% best22 / 52
72.7%29 / 52
LiveBench Long AgenticAgentic
47.5% best16 / 51
34.7%29 / 51
LiveBench Browser UseAgentic
88.0%8 / 49
92.0% best4 / 49