See key specs and per-benchmark scores for each model/mode. Scroll horizontally for all columns. 当前对比 2 个模型的评测数据与核心参数。

MiniMax M2.5
MiniMaxAI
Each axis is a category average, normalized to a 100-point radar.
Relative edge: AI Agent - 信息收集 +15.7 / Relative gap: 数学推理 -9.8
Relative edge: 数学推理 +9.8 / Relative gap: AI Agent - 信息收集 -15.7
Method: for each model and benchmark, the chart first averages all scores in the current mode scope instead of taking the best score, then averages those benchmark scores within each category. Only benchmarks with at least two selected models scored are included; missing values are not counted as zero.
Best overall
Kimi K2.5 · 63.18
Best single
Kimi K2.5 · AIME2025 96.10
Modality coverage
Kimi K2.5 · 2 modalities
Head to head
13
Benchmarks
7
Wins
6
Losses
-0.99
Average diff
Compare benchmark results across thinking modes and tool usage.
Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology
Complete scores for each model/mode across selected benchmarks.
13 benchmarks with comparable scores. Each model shows its best score; mode label is displayed below.
| Benchmark | MiniMax M2.5 | Kimi K2.5 |
|---|---|---|
ARC-AGI 综合评估 | 63.70Thinking Enabled | 65.30Thinking Enabled |
ARC-AGI-2 综合评估 | 4.90Thinking Enabled | 11.80Thinking Enabled |
GPQA Diamond 综合评估 | 85.20Thinking Enabled | 87.60Thinking Enabled |
HLE 综合评估 | 19.40Thinking Enabled | 50.20Thinking Enabled | Tools |
SWE-Bench Pro - Public 编程与软件工程 | 55.40Thinking Enabled | Tools | 50.70Thinking Enabled | Tools |
SWE-bench Verified 编程与软件工程 | 80.20Thinking Enabled | Tools | 76.80Thinking Enabled | Tools |
AIME2025 数学推理 | 86.30Thinking Enabled | 96.10Thinking Enabled |
BrowseComp AI Agent - 信息收集 | 76.30Thinking Enabled | Tools | 60.60Thinking Enabled | Tools |
Terminal Bench 2.0 AI Agent - 工具使用 | 51.70Thinking Enabled | Tools | 50.80Thinking Enabled | Tools |
GDPval-AA 生产力知识 | 36.00Thinking Enabled | 40.00Thinking Enabled |
AA-LCR 长上下文能力 | 69.50Thinking Enabled | 65.00Thinking Enabled |
Claw Bench OpenClaw智能体能力综合测评 | 92.10Thinking Enabled | Tools | 81.70Thinking Enabled | Tools |
Pinch Bench OpenClaw智能体能力综合测评 | 87.80Thinking Enabled | Tools | 84.80Thinking Enabled | Tools |
Side-by-side input/output token pricing
Licensing, MoE architecture, and multi-modality support.
| Features & specs | MiniMax M2.5MiniMaxAI | Kimi K2.5Moonshot AI |
|---|---|---|
Core specsRelease | 2026-02-12 | 2026-01-27 |
Context length | 128K | 256K |
Parameters | 2290 | 10000 |
Active parameters | 100 | 320 |
Max output | Not provided | 16384 |
MoE | Yes | Yes |
Supported modes | No mode data | 常规模式(Non-Thinking Mode)思考模式(Thinking Mode) |
LicenseCode Open Source | Closed Source | Not provided |
Weights Open Source | Not provided | Not provided |
Commercial use | 免费商用授权 | 免费商用授权 |
Modality supportText Input/Output | Not provided | / |
Image Input/Output | Not provided | / |
ResourcesPaper / report | MiniMax M2.5: Built for Real-World Productivity. | Kimi K2.5: Visual Agentic Intelligence |
DataLearner blog | Not provided | 重磅!Kimi K2.5发布,依然免费开源!原生多模态MoE架构,全球最大规模参数的开源模型之一,官方评测结果比肩诸多闭源模型!可以驱动100个子Agent执行! |

Kimi K2.5
Moonshot AI