MiniMax M2.5vsM2.1
Across 10 shared benchmarks, MiniMax M2.5 leads overall: MiniMax M2.5 wins 8, M2.1 wins 1, with 1 ties and an average score difference of +8.21.
MiniMax M2.5
MiniMaxAI · 2026-02-12 · Reasoning model
M2.1
MiniMaxAI · 2025-12-23 · Chat model
MiniMax M2.58 wins(80%)Ties1(10%)1 winM2.1
Benchmark scores
Grouped by capability, sorted by largest gap within each. 10 shared benchmarks.
Coding and Software Engineer
MiniMax M2.5 2/2| Benchmark | MiniMax M2.5 | M2.1 | Diff |
|---|---|---|---|
| SWE-Bench Pro - Public | 55.4018 / 43 | 32.6042 / 43 | +22.80 |
| SWE-bench Verified | 80.2013 / 108 | 74.8035 / 108 | +5.40 |
General Knowledge
Even 2/2| Benchmark | MiniMax M2.5 | M2.1 | Diff |
|---|---|---|---|
| GPQA Diamond | 85.2048 / 178Thinking (No Tools) | 8169 / 178 | +4.20 |
| HLE | 19.40106 / 157Thinking (No Tools) | 2294 / 157 | -2.60 |
Agent Level Benchmark
MiniMax M2.5 1/1| Benchmark | MiniMax M2.5 | M2.1 | Diff |
|---|---|---|---|
| τ²-Bench - Telecom | 97.8010 / 35 | 8722 / 35 | +10.80 |
AI Agent - Information Search
MiniMax M2.5 1/1| Benchmark | MiniMax M2.5 | M2.1 | Diff |
|---|---|---|---|
| BrowseComp | 76.3018 / 45 | 47.4037 / 45 | +28.90 |
AI Agent - Tool Usage
MiniMax M2.5 1/1| Benchmark | MiniMax M2.5 | M2.1 | Diff |
|---|---|---|---|
| Terminal Bench 2.0 | 51.7030 / 46 | 47.9035 / 46 | +3.80 |
Claw-style Agent Evaluation
MiniMax M2.5 1/1| Benchmark | MiniMax M2.5 | M2.1 | Diff |
|---|---|---|---|
| Pinch Bench | 87.806 / 37Thinking (With Tools) | 84.3018 / 37Thinking (With Tools) | +3.50 |
Instruction Following
Even 1/1| Benchmark | MiniMax M2.5 | M2.1 | Diff |
|---|---|---|---|
| IF Bench | 7012 / 29 | 7012 / 29 | — |
Math and Reasoning
MiniMax M2.5 1/1| Benchmark | MiniMax M2.5 | M2.1 | Diff |
|---|---|---|---|
| AIME2025 | 86.3048 / 106Thinking (No Tools) | 8156 / 106 | +5.30 |
Specs
| Field | MiniMax M2.5 | M2.1 |
|---|---|---|
| Publisher | MiniMaxAI | MiniMaxAI |
| Release date | 2026-02-12 | 2025-12-23 |
| Model type | Reasoning model | Chat model |
| Architecture | MoE | MoE |
| Parameters | 229B | 230B |
| Context length | 128K | 200K |
| Max output | Not available | 128K |
API pricing
Prices use DataLearner records when available; missing fields are not inferred.
| Item | MiniMax M2.5 | M2.1 |
|---|---|---|
| Text input | $0.3 / 1M tokens | Not public |
| Text output | $2.4 / 1M tokens | Not public |
One or both models have incomplete public pricing.
Summary
- MiniMax M2.5leads in:Coding and Software Engineer (2/2), Agent Level Benchmark (1/1), AI Agent - Information Search (1/1), AI Agent - Tool Usage (1/1), Claw-style Agent Evaluation (1/1), Math and Reasoning (1/1)
- Tied in:General Knowledge, Instruction Following
On average across the 10 shared benchmarks, MiniMax M2.5 scores 8.21 higher.
Largest single-benchmark gap: BrowseComp — MiniMax M2.5 76.30 vs M2.1 47.40 (+28.90).
Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.