加载中...
加载中...
See key specs and per-benchmark scores for each model/mode. Scroll horizontally for all columns. 当前对比 4 个模型的评测数据与核心参数。
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
Performance benchmarks
Compare benchmark results across thinking modes and tool usage.
Best Overall
Kimi K2 0905 · 52.65
Best Single
DeepSeek-V3.1 · AIME2025 88.40
Thinking Mode (Default)
Kimi K2 0905 · 1 All Modes
Higher is usually better; “—” means no score.
Complete scores for each model/mode across selected benchmarks.
Higher is usually better; “—” means no score.
| Benchmark | KI Kimi K2 0905Moonshot AI | DE DeepSeek-V3.1DeepSeek-AI | QW Qwen3-Coder-480B-A35B阿里巴巴 |
|---|---|---|---|
| 综合评估 | |||
HLE | 21.70 | 15.90 | — |
| 编程与软件工程 | |||
SWE-bench Verified | 69.20 | — | 67.00 |
| 数学推理 | |||
AIME2025 | 75.20 | 88.40 | — |
| AI Agent - 工具使用 | |||
Terminal-Bench | 44.50 | — | 37.50 |
Feature compare
Licensing, MoE architecture, and multi-modality support.
| Features & specs | KI Kimi K2 0905Moonshot AI | KI Kimi K2Moonshot AI | DE DeepSeek-V3.1DeepSeek-AI | QW Qwen3-Coder-480B-A35B阿里巴巴 |
|---|---|---|---|---|
Model snapshots | ||||
Organization | Moonshot AI | Moonshot AI | DeepSeek-AI | 阿里巴巴 |
模型全名 | Kimi K2-Instruct-0905 | Kimi-K2-0711-Preview | DeepSeek-V3.1 | Qwen3-Coder-480B-A35B-Instruct |
模型简介 | Not provided | Not provided | Not provided | Not provided |
模型类型 | 聊天大模型 | 聊天大模型 | 聊天大模型 | 编程大模型 |
模型代号 | kimi-k2-0905 | kimi-k2-0711-base-preview | deepseek-v-3_1 | Qwen3-Coder-480B-A35B-Instruct |
Release | 2025-09-05 | 2025-07-11 | 2025-08-20 | 2025-07-23 |
MoE | Yes | Yes | Yes | Yes |
规格与性能 | ||||
Context length | 256K | 131K | 128K | 256K |
Parameters | 10000 | 10000 | 6710 | 4800 |
激活参数量 | 320 | 320 | 370 | 350 |
模型规模 | 100b | 100b | 100b | 100b |
模型大小 | 1.01TB | 1.01TB | 1340GB | 470.77 GB |
推理速度 | ||||
推理等级 | ||||
最大输出 | 4096 | 134144 | 8192 | 16384 |
Supported modes | 常规模式（Non-Thinking Mode） | 常规模式（Non-Thinking Mode） | 常规模式（Non-Thinking Mode）思考模式（Thinking Mode） | 常规模式（Non-Thinking Mode） |
开源与许可 | ||||
Code Open Source | Closed Source | Not provided | Closed Source | Not provided |
Weights Open Source | Closed Source | Not provided | Closed Source | Not provided |
Commercial use | 免费商用授权 | 免费商用授权 | 免费商用授权 | 免费商用授权 |
Modality support | ||||
Text Input/Output | / | / | / | / |
Image Input/Output | / | / | / | / |
Audio Input/Output | / | / | / | / |
Video Input/Output | / | / | / | / |
Embedding Input/Output | / | / | / | / |
API 接口详情 | ||||
Text 价格 | Input: 0.60 美元/ 100 万tokensOutput: 2.5 美元/ 100 万tokens | Input: 0.6 美元/100 万tokensOutput: 2.5 美元/100 万tokens | Input: 0.56 美元/100 万tokensOutput: 1.68 美元/100 万tokens | Not provided |
Image API pricing | Not provided | Not provided | Not provided | Not provided |
Audio API pricing | Not provided | Not provided | Not provided | Not provided |
Video API pricing | Not provided | Not provided | Not provided | Not provided |
Embedding API pricing | Not provided | Not provided | Not provided | Not provided |
Resources | ||||
GitHub | Not provided | Repo | Not provided | Repo |
Hugging Face | Model Page | Model Page | Model Page | Model Page |
Official Page | Not provided | Not provided | Not provided | Not provided |
Guides | Not provided | Not provided | Not provided | Not provided |
Papers | Kimi K2: Open Agentic Intelligence | DeepSeek-V3.1 Release | Qwen3-Coder: Agentic Coding in the World | |
DataLearnerAI | Moonshot AI发布Kimi K2-Instruct-0905：256K上下文长度加持，全面升级的开放式智能体模型 | Kimi开源K2大模型：全球首个开源可商用的1万亿参数规模大模型，MoE架构，评测结果与DeepSeekV3相当，但模型文件有1TB！ | DeepSeek V4没有等到，但是DeepSeekAI把DeepSeek V3升级到DeepSeek V3.1了，小幅更新，但核心架构和参数不变 | 阿里开源全新编程大模型Qwen3-Coder-480B-A35B，官方宣称其编程水平接近Claude Sonnet 4，免费开源可商用，同时开源Claude Code免费平替选择Qwen Code |
API pricing
Side-by-side input/output token pricing