Inference
EN

MiMo-V2.6-Pro vs Step 5 Preview

Change models
MiMo-V2.6-ProXiaomi
Step 5 PreviewStepFun
IndexAll categories, out of 100.
70.2 ±4.7 best
68.4 ±3.7
GapIndex points below the top.
Top
−1.8 · no clear gap
CoverageIndex weight that has results.
80%
90% best
SpeedOutput tokens per second.
31/s
105/s best
EloLMArena rating and rank.
1492 best
1443
ReleasedMonth of release.
2026-09
2026-09
Input / 1MUS dollars per 1M input tokens.
$0.435 best
$1.00
Output / 1MUS dollars per 1M output tokens.
$0.87 best
$2.70
1K chats2K tokens in, 500 out, each.
$1.31 best
$3.35
1K documents20K tokens in, 1K out, each.
$9.57 best
$22.70
1K agent tasks200K tokens in, 20K out, each.
$104.40 best
$254.00
ContextMaximum tokens in one request.
1.05M best
1M
AgenticMulti-step tasks with tools.
72.1 best
68.8
CodingCode writing and repair.
66.6
71.4 best
ReasoningLogic problems and puzzles.
68.3 best
67.4
MultimodalTasks with images and text.
—
62.2
KnowledgeFacts and expert knowledge.
64.2 best
63.5
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
—
—
MathMath problems.
71.3 best
61.7

Benchmarks

43 shared
Only MiMo-V2.6-Pro: 15Only Step 5 Preview: 19
MiMo-V2.6-Pro
Step 5 Preview
Artificial Analysis Humanity's Last ExamKnowledge
49.4% best14 / 222
46.5%23 / 222
Artificial Analysis Long Context ReasoningReasoning
86.3%3 / 217
88.3% best2 / 217
Critical Physics TasksReasoning
26.6% best19 / 214
20.9%26 / 214
Artificial Analysis Omniscience AccuracyKnowledge
34.8%67 / 164
36.4% best64 / 164
GDPval-AA normalizedAgentic
59.2% best9 / 133
54.3%24 / 133
Vibe Code Bench v1.1Coding
85.2% best16 / 123
70.4%48 / 123
Artificial Analysis SciCodeCoding
60.9% best8 / 114
58.9%13 / 114
Legal Research BenchLegal
47.1% best16 / 90
40.4%35 / 90
Terminal-Bench 2.1 · BenchLMAgentic
89.9% best4 / 86
85.0%20 / 86
Tax Agent BenchFinance
32.4% best22 / 82
27.3%39 / 82
LiveBench Data AnalysisReasoning
72.3% best49 / 80
63.5%69 / 80
LiveBench ReasoningReasoning
79.6%59 / 80
83.8% best44 / 80
Finance Agent (v2)Finance
57.3% best16 / 79
50.7%42 / 79
Code MigrationCoding
43.0% best26 / 77
39.0%31 / 77
EMBFinance
62.9% best30 / 75
57.1%41 / 75
Terminal-Bench 4.0Agentic
31.3%24 / 61
31.8% best23 / 61
DeepSWEAgentic
71.9% best18 / 59
67.7%27 / 59
Agent Arena steerabilityAgentic
1.5 best25 / 54
−0.930 / 54
Agent Arena task outcomeAgentic
5.0 best16 / 54
−1.535 / 54
ProofBench v1.1Math
70.0% best21 / 54
42.0%37 / 54