DataLearner logo

MiniMax-M2.7vsMiniMax M2.5

Across 11 shared benchmarks, MiniMax-M2.7 leads overall: MiniMax-M2.7 wins 7, MiniMax M2.5 wins 4, with 0 ties and an average score difference of +2.20.

MiniMaxAI
MiniMax-M2.7

MiniMaxAI · 2026-03-18 · Reasoning model

MiniMaxAI
MiniMax M2.5

MiniMaxAI · 2026-02-12 · Reasoning model

MiniMax-M2.77 wins(64%)(36%)4 winsMiniMax M2.5

Benchmark scores

Grouped by capability, sorted by largest gap within each. 11 shared benchmarks.

General Knowledge

MiniMax-M2.7 2/3
BenchmarkMiniMax-M2.7MiniMax M2.5Diff
HLE29.60208 / 563Thinking (No Tools) · Text only20.50281 / 563Thinking (No Tools) · Text only+9.10
LiveBench63.4957 / 117Deep Thinking (No Tools)60.1470 / 117Deep Thinking (No Tools)+3.35
CritPt0.60168 / 200Thinking (No Tools)1.10147 / 200Thinking (No Tools)-0.50

Agent Level Benchmark

Even 2/2
BenchmarkMiniMax-M2.7MiniMax M2.5Diff
τ²-Bench - Telecom8589 / 264Thinking (With Tools)97.8012 / 264Thinking (With Tools)-12.80
Terminal Bench Hard3955 / 244Thinking (With Tools)34.8075 / 244Thinking (With Tools)+4.20

Claw-style Agent Evaluation

MiniMax M2.5 2/2
BenchmarkMiniMax-M2.7MiniMax M2.5Diff
Pinch Bench87.1010 / 38Thinking (With Tools)87.807 / 38Thinking (With Tools)-0.70
Claw Bench91.705 / 29Thinking (With Tools)92.104 / 29Thinking (With Tools)-0.40

AI Agent - Tool Usage

MiniMax-M2.7 1/1
BenchmarkMiniMax-M2.7MiniMax M2.5Diff
Terminal Bench 2.05725 / 48Thinking (With Tools)51.7032 / 48Thinking (With Tools)+5.30

Coding and Software Engineer

MiniMax-M2.7 1/1
BenchmarkMiniMax-M2.7MiniMax M2.5Diff
SWE-Bench Pro - Public56.2029 / 62Thinking (With Tools)55.4031 / 62Thinking (With Tools)+0.80

General Evaluation

MiniMax-M2.7 1/1
BenchmarkMiniMax-M2.7MiniMax M2.5Diff
GPQA Diamond87133 / 462Thinking (No Tools)85.20157 / 462Thinking (No Tools)+1.80

Productivity Knowledge

MiniMax-M2.7 1/1
BenchmarkMiniMax-M2.7MiniMax M2.5Diff
GDPval-AA507 / 15Thinking (No Tools)3611 / 15Thinking (No Tools)+14

Specs

FieldMiniMax-M2.7MiniMax M2.5
PublisherMiniMaxAIMiniMaxAI
Release date2026-03-182026-02-12
Model typeReasoning modelReasoning model
ArchitectureMoEMoE
Parameters229B229B
Context length200K128K
Max output200KNot available

API pricing

Prices use DataLearner records when available; missing fields are not inferred.

ItemMiniMax-M2.7MiniMax M2.5
Text input$0.3 / 1M tokens$0.3 / 1M tokens
Text output$1.2 / 1M tokens$2.4 / 1M tokens
Cache read$0.06 / 1M tokensNot public
Cache write$0.375 / 1M tokensNot public

Summary

  • MiniMax-M2.7leads in:General Knowledge (2/3), AI Agent - Tool Usage (1/1), Coding and Software Engineer (1/1), General Evaluation (1/1), Productivity Knowledge (1/1)
  • MiniMax M2.5leads in:Claw-style Agent Evaluation (2/2)
  • Tied in:Agent Level Benchmark

On average across the 11 shared benchmarks, MiniMax-M2.7 scores 2.20 higher.

Largest single-benchmark gap: GDPval-AA — MiniMax-M2.7 50 vs MiniMax M2.5 36 (+14).

Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.