Inference
EN

Hy4 vs MiMo-V2.6-Pro

Change models
Hy4Tencent
MiMo-V2.6-ProXiaomi
IndexAll categories, out of 100.
72.0 ±7.8 best
70.2 ±4.7
GapIndex points below the top.
Top
−1.8 · no clear gap
CoverageIndex weight that has results.
65%
80% best
SpeedOutput tokens per second.
52/s best
31/s
EloLMArena rating and rank.
—
1492
ReleasedMonth of release.
2026-11
2026-09
Input / 1MUS dollars per 1M input tokens.
$0.417 best
$0.435
Output / 1MUS dollars per 1M output tokens.
$1.25
$0.87 best
1K chats2K tokens in, 500 out, each.
$1.46
$1.31 best
1K documents20K tokens in, 1K out, each.
$9.59
$9.57 best
1K agent tasks200K tokens in, 20K out, each.
$108.41
$104.40 best
ContextMaximum tokens in one request.
1.05M
1.05M
AgenticMulti-step tasks with tools.
74.5 best
72.1
CodingCode writing and repair.
68.4 best
66.6
ReasoningLogic problems and puzzles.
—
68.3
MultimodalTasks with images and text.
57.4
—
KnowledgeFacts and expert knowledge.
83.7 best
64.2
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
—
—
MathMath problems.
—
71.3

Benchmarks

11 shared
Only Hy4: 10Only MiMo-V2.6-Pro: 47
Hy4
MiMo-V2.6-Pro
Vibe Code Bench v1.1Coding
81.0%30 / 123
85.2% best16 / 123
Legal Research BenchLegal
41.3%30 / 90
47.1% best16 / 90
Tax Agent BenchFinance
34.4% best17 / 82
32.4%22 / 82
Terminal-Bench 4.0Agentic
39.4% best15 / 61
31.3%24 / 61
DeepSWEAgentic
74.9% best8 / 59
71.9%18 / 59
Agent Arena steerabilityAgentic
8.8 best14 / 54
1.525 / 54
Agent Arena task outcomeAgentic
6.7 best12 / 54
5.016 / 54
HLAB-Vals-1.1Legal
12.5%17 / 35
13.8% best13 / 35
AutomationBenchAgentic
56.2% best4 / 32
53.1%7 / 32
Agents' Last ExamAgentic
41.0% best8 / 21
31.6%12 / 21
ProgramBench: Can Language Models Rebuild Programs From Scratch?Coding
30.0% best12 / 19
26.5%15 / 19