MiniMax M2.5vsKimi K2.5
Across 13 shared benchmarks, MiniMax M2.5 leads overall: MiniMax M2.5 wins 7, Kimi K2.5 wins 6, with 0 ties and an average score difference of -0.99.
MiniMax M2.5
MiniMaxAI · 2026-02-12 · Reasoning model
Kimi K2.5
Moonshot AI · 2026-01-27 · Multimodal model
MiniMax M2.57 wins(54%)(46%)6 winsKimi K2.5
Benchmark scores
Grouped by capability, sorted by largest gap within each. 13 shared benchmarks.
General Knowledge
Kimi K2.5 4/4| Benchmark | MiniMax M2.5 | Kimi K2.5 | Diff |
|---|---|---|---|
| HLE | 19.4098 / 149Thinking (No Tools) | 50.2017 / 149Thinking (With Tools) | -30.80 |
| ARC-AGI-2 | 4.9043 / 58Thinking (No Tools) | 11.8035 / 58Thinking (No Tools) | -6.90 |
| GPQA Diamond | 85.2045 / 175Thinking (No Tools) | 87.6031 / 175Thinking (No Tools) | -2.40 |
| ARC-AGI | 63.7032 / 65Thinking (No Tools) | 65.3031 / 65Thinking (No Tools) | -1.60 |
Claw-style Agent Evaluation
MiniMax M2.5 2/2| Benchmark | MiniMax M2.5 | Kimi K2.5 | Diff |
|---|---|---|---|
| Claw Bench | 92.104 / 29Thinking (With Tools) | 81.7018 / 29Thinking (With Tools) | +10.40 |
| Pinch Bench | 87.806 / 37Thinking (With Tools) | 84.8017 / 37Thinking (With Tools) | +3 |
Coding and Software Engineer
MiniMax M2.5 2/2| Benchmark | MiniMax M2.5 | Kimi K2.5 | Diff |
|---|---|---|---|
| SWE-Bench Pro - Public | 55.4013 / 36thinking + 使用工具 | 50.7025 / 36Thinking (With Tools) | +4.70 |
| SWE-bench Verified | 80.209 / 103thinking + 使用工具 | 76.8022 / 103Thinking (With Tools) | +3.40 |
AI Agent - Information Search
MiniMax M2.5 1/1| Benchmark | MiniMax M2.5 | Kimi K2.5 | Diff |
|---|---|---|---|
| BrowseComp | 76.3016 / 43thinking + 使用工具 | 60.6027 / 43Thinking (With Tools + Internet) | +15.70 |
AI Agent - Tool Usage
MiniMax M2.5 1/1| Benchmark | MiniMax M2.5 | Kimi K2.5 | Diff |
|---|---|---|---|
| Terminal Bench 2.0 | 51.7027 / 43thinking + 使用工具 | 50.8030 / 43Thinking (With Tools) | +0.90 |
Long Context
MiniMax M2.5 1/1| Benchmark | MiniMax M2.5 | Kimi K2.5 | Diff |
|---|---|---|---|
| AA-LCR | 69.503 / 13Thinking (No Tools) | 6510 / 13Thinking (No Tools) | +4.50 |
Math and Reasoning
Kimi K2.5 1/1| Benchmark | MiniMax M2.5 | Kimi K2.5 | Diff |
|---|---|---|---|
| AIME2025 | 86.3048 / 106Thinking (No Tools) | 96.1021 / 106Thinking (No Tools) | -9.80 |
Productivity Knowledge
Kimi K2.5 1/1| Benchmark | MiniMax M2.5 | Kimi K2.5 | Diff |
|---|---|---|---|
| GDPval-AA | 3616 / 20Thinking (No Tools) | 4014 / 20Thinking (No Tools) | -4 |
Specs
| Field | MiniMax M2.5 | Kimi K2.5 |
|---|---|---|
| Publisher | MiniMaxAI | Moonshot AI |
| Release date | 2026-02-12 | 2026-01-27 |
| Model type | Reasoning model | Multimodal model |
| Architecture | MoE | MoE |
| Parameters | 2290.0 | 10000.0 |
| Context length | 128K | 256K |
| Max output | Not available | 16384 |
API pricing
Prices use DataLearner records when available; missing fields are not inferred.
| Item | MiniMax M2.5 | Kimi K2.5 |
|---|---|---|
| Text input | $0.3 / 1M tokens | 0.6 美元/100 万tokens |
| Text output | $2.4 / 1M tokens | 3 美元/100 万tokens |
| Cache read | Not public | 0.1 美元/100 万tokens |
Summary
- MiniMax M2.5leads in:Claw-style Agent Evaluation (2/2), Coding and Software Engineer (2/2), AI Agent - Information Search (1/1), AI Agent - Tool Usage (1/1), Long Context (1/1)
- Kimi K2.5leads in:General Knowledge (4/4), Math and Reasoning (1/1), Productivity Knowledge (1/1)
On average across the 13 shared benchmarks, Kimi K2.5 scores 0.99 higher.
Largest single-benchmark gap: HLE — MiniMax M2.5 19.40 vs Kimi K2.5 50.20 (-30.80).
Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.