Compare · ModelsLive · 2 picked · head to head
MiniMax M2.5 vs MiniMax M2.7
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
MiniMax M2.7 wins on 9/12 benchmarks
MiniMax M2.7 wins 9 of 12 shared benchmarks. Leads in arena · reasoning · language.
Category leads
arena·MiniMax M2.7coding·MiniMax M2.5reasoning·MiniMax M2.7language·MiniMax M2.7math·MiniMax M2.7knowledge·MiniMax M2.7general·MiniMax M2.5
Hype vs Reality
Attention vs performance
MiniMax M2.5
#131 by perf·#11 by attention
MiniMax M2.7
#120 by perf·#11 by attention
Best value
MiniMax M2.7
1.3x better value than MiniMax M2.5
MiniMax M2.5
72.0 pts/$
$0.68/M
MiniMax M2.7
96.2 pts/$
$0.53/M
Vendor risk
Mixed exposure
One or more vendors flagged
MiniMax
$4.0B·Tier 1
MiniMax
$4.0B·Tier 1
Head to head
12 benchmarks · 2 models
MiniMax M2.5MiniMax M2.7
Chatbot Arena Elo · Coding
MiniMax M2.7 leads by +10.7
MiniMax M2.5
1386.3
MiniMax M2.7
1397.0
Chatbot Arena Elo · Overall
MiniMax M2.7 leads by +24.1
MiniMax M2.5
1390.8
MiniMax M2.7
1415.0
LiveBench · Agentic Coding
MiniMax M2.5 leads by +1.7
MiniMax M2.5
51.7
MiniMax M2.7
50.0
LiveBench · Coding
MiniMax M2.5 leads by +15.8
MiniMax M2.5
70.7
MiniMax M2.7
54.9
LiveBench · Data Analysis
MiniMax M2.7 leads by +6.7
MiniMax M2.5
49.6
MiniMax M2.7
56.3
LiveBench · If
MiniMax M2.7 leads by +3.9
MiniMax M2.5
57.2
MiniMax M2.7
61.1
LiveBench · Language
MiniMax M2.7 leads by +11.7
MiniMax M2.5
55.1
MiniMax M2.7
66.8
LiveBench · Mathematics
MiniMax M2.7 leads by +3.1
MiniMax M2.5
77.4
MiniMax M2.7
80.5
LiveBench · Overall
MiniMax M2.7 leads by +3.4
MiniMax M2.5
60.1
MiniMax M2.7
63.5
LiveBench · Reasoning
MiniMax M2.7 leads by +15.5
MiniMax M2.5
59.3
MiniMax M2.7
74.8
Proofbench
MiniMax M2.5 leads by +1.0
MiniMax M2.5
4.0
MiniMax M2.7
3.0
Terminal Bench
MiniMax M2.7 leads by +2.4
Terminal-Bench 2.0 · evaluates AI agents on real terminal-based coding tasks · writing scripts, debugging, running tests, and managing projects entirely through command-line interaction. Tests both code quality and terminal fluency. Claude Opus 4.7 scores 69.4%, demonstrating significant agentic terminal competence.
MiniMax M2.5
42.7
MiniMax M2.7
45.1
Full benchmark table
| Benchmark | MiniMax M2.5 | MiniMax M2.7 |
|---|---|---|
Chatbot Arena Elo · Coding | 1386.3 | 1397.0 |
Chatbot Arena Elo · Overall | 1390.8 | 1415.0 |
LiveBench · Agentic Coding | 51.7 | 50.0 |
LiveBench · Coding | 70.7 | 54.9 |
LiveBench · Data Analysis | 49.6 | 56.3 |
LiveBench · If | 57.2 | 61.1 |
LiveBench · Language | 55.1 | 66.8 |
LiveBench · Mathematics | 77.4 | 80.5 |
LiveBench · Overall | 60.1 | 63.5 |
LiveBench · Reasoning | 59.3 | 74.8 |
Proofbench | 4.0 | 3.0 |
Terminal Bench Terminal-Bench 2.0 · evaluates AI agents on real terminal-based coding tasks · writing scripts, debugging, running tests, and managing projects entirely through command-line interaction. Tests both code quality and terminal fluency. Claude Opus 4.7 scores 69.4%, demonstrating significant agentic terminal competence. | 42.7 | 45.1 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $0.27 | $1.08 | 205K tokens (~102 books) | $4.72 | |
| $0.21 | $0.84 | 205K tokens (~102 books) | $3.67 |