Inference
EN

MiMo-V2.6-Flash vs MiMo-V2.6-Pro

Change models
MiMo-V2.6-FlashXiaomi
MiMo-V2.6-ProXiaomi
IndexAll categories, out of 100.
64.3 ±4.4
70.2 ±4.7 best
GapIndex points below the top.
−5.9 · no clear gap
Top
CoverageIndex weight that has results.
90% best
80%
SpeedOutput tokens per second.
28/s
31/s best
EloLMArena rating and rank.
1450
1492 best
ReleasedMonth of release.
2026-09
2026-09
Input / 1MUS dollars per 1M input tokens.
$0.14 best
$0.435
Output / 1MUS dollars per 1M output tokens.
$0.28 best
$0.87
1K chats2K tokens in, 500 out, each.
$0.42 best
$1.31
1K documents20K tokens in, 1K out, each.
$3.08 best
$9.57
1K agent tasks200K tokens in, 20K out, each.
$33.60 best
$104.40
ContextMaximum tokens in one request.
1.05M
1.05M
AgenticMulti-step tasks with tools.
69.7
72.1 best
CodingCode writing and repair.
64.1
66.6 best
ReasoningLogic problems and puzzles.
59.2
68.3 best
MultimodalTasks with images and text.
50.3
—
KnowledgeFacts and expert knowledge.
52.3
64.2 best
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
—
—
MathMath problems.
68.9
71.3 best

Benchmarks

39 shared
Only MiMo-V2.6-Flash: 2Only MiMo-V2.6-Pro: 19
MiMo-V2.6-Flash
MiMo-V2.6-Pro
Artificial Analysis Humanity's Last ExamKnowledge
35.1%62 / 222
49.4% best14 / 222
Artificial Analysis Long Context ReasoningReasoning
74.3%98 / 217
86.3% best3 / 217
Critical Physics TasksReasoning
12.0%57 / 214
26.6% best19 / 214
Artificial Analysis Omniscience AccuracyKnowledge
27.2%91 / 164
34.8% best67 / 164
GDPval-AA normalizedAgentic
55.4%19 / 133
59.2% best9 / 133
Vibe Code Bench v1.1Coding
79.0%33 / 123
85.2% best16 / 123
Artificial Analysis SciCodeCoding
51.3%46 / 114
60.9% best8 / 114
MedScribeHealthcare
85.3%29 / 109
88.3% best13 / 109
MedCodeHealthcare
41.1%58 / 106
45.0% best39 / 106
Legal Research BenchLegal
38.0%43 / 90
47.1% best16 / 90
Terminal-Bench 2.1 · BenchLMAgentic
87.6%13 / 86
89.9% best4 / 86
Tax Agent BenchFinance
28.5%37 / 82
32.4% best22 / 82
Finance Agent (v2)Finance
56.3%19 / 79
57.3% best16 / 79
Code MigrationCoding
40.9%29 / 77
43.0% best26 / 77
EMBFinance
65.5% best24 / 75
62.9%30 / 75
Terminal-Bench 2.1 · Vals AIAgentic
76.4% best16 / 75
67.8%32 / 75
ProgramBenchCoding
0.5%19 / 65
0.5%19 / 65
Terminal-Bench 4.0Agentic
24.2%30 / 61
31.3% best24 / 61
DeepSWEAgentic
67.9%26 / 59
71.9% best18 / 59
Agent Arena steerabilityAgentic
2.8 best20 / 54
1.525 / 54