MiniMax M2.7 vs MiniMax M3

Change models
MiniMax M3MiniMax
IndexAll categories, out of 100.
51.2 ±4.5
55.7 ±2.9 best
GapIndex points below the top.
−4.5 · no clear gap
Top
CoverageIndex weight that has results.
85%
95% best
SpeedOutput tokens per second.
35/s
79/s best
EloLMArena rating and rank.
1405
1428 best
ReleasedMonth of release.
2026-03
2026-06
Input / 1MUS dollars per 1M input tokens.
$0.21 best
$0.30
Output / 1MUS dollars per 1M output tokens.
$0.84 best
$1.20
Batch output / 1MBatch price per 1M output tokens.
—
$0.88
1K chats2K tokens in, 500 out, each.
$0.84 best
$1.20
1K documents20K tokens in, 1K out, each.
$5.04 best
$7.20
1K agent tasks200K tokens in, 20K out, each.
$58.80 best
$84.00
ContextMaximum tokens in one request.
197K
1.05M best
AgenticMulti-step tasks with tools.
52.0
57.4 best
CodingCode writing and repair.
52.3
57.7 best
ReasoningLogic problems and puzzles.
49.5
50.5 best
MultimodalTasks with images and text.
—
56.5
KnowledgeFacts and expert knowledge.
50.2
61.0 best
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
59.8 best
54.7
MathMath problems.
47.7 best
41.6

Benchmarks

29 shared
Only MiniMax M2.7: 24Only MiniMax M3: 55
MiniMax M2.7
MiniMax M3
29.6%82 / 222
39.0% best52 / 222
78.3%63 / 217
83.0% best18 / 217
0.6%132 / 214
3.7% best93 / 214
87.4%57 / 199
92.9% best13 / 199
26.8%92 / 164
44.9% best38 / 164
84.0%43 / 149
85.4% best21 / 149
84.8%58 / 149
88.9% best45 / 149
75.7%19 / 146
82.9% best1 / 146
TaxEval v2Finance
66.6%110 / 143
72.7% best58 / 143
79.9%71 / 142
82.2% best53 / 142
86.6%41 / 135
92.7% best14 / 135
MMLU ProKnowledge
80.4%88 / 135
84.2% best64 / 135
25.5%72 / 133
37.1% best48 / 133
CorpFin v2Finance
61.2%55 / 131
68.1% best11 / 131
11.9%103 / 123
47.6% best72 / 123
50.1% best52 / 114
47.1%65 / 114
MedScribeHealthcare
79.9%60 / 109
87.3% best18 / 109
MedCodeHealthcare
34.4%79 / 106
46.3% best37 / 106
10.6%79 / 90
29.8% best56 / 90
SWE-benchCoding
73.8%51 / 87
75.0% best44 / 87