DeepSeek-V4.1-FlashvsKimi K3
在 15 个共同 benchmark 中，DeepSeek-V4.1-Flash 整体领先：DeepSeek-V4.1-Flash 领先 13 项，Kimi K3 领先 1 项，持平 1 项，平均分差 +6.35。
DeepSeek-V4.1-Flash
DeepSeek-AI · 2026-09-10 · 多模态大模型
Kimi K3
Moonshot AI · 2026-07-16 · 推理大模型
DeepSeek-V4.1-Flash13 项(87%)持平1(7%)1 项Kimi K3
评测分数
按能力类目分组，每组内按分差大小排列；共 15 项。
AI Agent - Tool Usage
DeepSeek-V4.1-Flash 领先 4/4| 评测项 | DeepSeek-V4.1-Flash | Kimi K3 | 分差 |
|---|---|---|---|
| Terminal-Bench 4.0 | 31.207 / 16Max (With Tools) | 12.6013 / 16Max (With Tools) | +18.60 |
| Terminal-Bench 3.0 | 304 / 11Max (With Tools) | 17.708 / 11Max (With Tools) | +12.30 |
| CyberGym | 88.101 / 8Max (With Tools) | 805 / 8Max (With Tools) | +8.10 |
| Terminal-Bench 2.1 | 90.601 / 53Max (With Tools) | 88.307 / 53Max (With Tools) | +2.30 |
Coding and Software Engineer
DeepSeek-V4.1-Flash 领先 3/3| 评测项 | DeepSeek-V4.1-Flash | Kimi K3 | 分差 |
|---|---|---|---|
| NL2Repo-Bench | 65.402 / 16Max (With Tools) | 586 / 16Max (With Tools) | +7.40 |
| DeepSWE | 74.202 / 38Max (With Tools) | 67.5011 / 38Max (With Tools) | +6.70 |
| Program Bench | 20.308 / 11Max (With Tools) | 17.5010 / 11Max (With Tools) | +2.80 |
Multimodal Understanding
DeepSeek-V4.1-Flash 领先 3/3| 评测项 | DeepSeek-V4.1-Flash | Kimi K3 | 分差 |
|---|---|---|---|
| Chartography | 78.904 / 8Max (With Tools) | 68.106 / 8Max (With Tools) | +10.80 |
| ZeroBench Main | 493 / 6Max (With Tools) | 414 / 6Max (With Tools) | +8 |
| BabyVision | 89.603 / 9Max (With Tools) | 85.705 / 9Max (With Tools) | +3.90 |
Agent Level Benchmark
DeepSeek-V4.1-Flash 领先 1/1| 评测项 | DeepSeek-V4.1-Flash | Kimi K3 | 分差 |
|---|---|---|---|
| Agents' Last Exam | 31.806 / 19Max (With Tools) | 27.609 / 19Max (With Tools) | +4.20 |
General Knowledge
DeepSeek-V4.1-Flash 领先 1/1| 评测项 | DeepSeek-V4.1-Flash | Kimi K3 | 分差 |
|---|---|---|---|
| HLE | 63.903 / 197Max (With Tools) | 59.808 / 197Max (With Tools) | +4.10 |
Math and Reasoning
胶着 1/1| 评测项 | DeepSeek-V4.1-Flash | Kimi K3 | 分差 |
|---|---|---|---|
| MathArena Apex | 65.601 / 3Max (No Tools) | 65.601 / 3Max (No Tools) | 持平 |
Productivity Knowledge
DeepSeek-V4.1-Flash 领先 1/1| 评测项 | DeepSeek-V4.1-Flash | Kimi K3 | 分差 |
|---|---|---|---|
| AutomationBench | 54.801 / 17Max (With Tools) | 46.707 / 17Max (With Tools) | +8.10 |
科学与综合推理
Kimi K3 领先 1/1| 评测项 | DeepSeek-V4.1-Flash | Kimi K3 | 分差 |
|---|---|---|---|
| GPQA Diamond | 90.9039 / 274Max (No Tools) | 92.9023 / 274Max (No Tools) | -2 |
规格对比
| 字段 | DeepSeek-V4.1-Flash | Kimi K3 |
|---|---|---|
| 发布机构 | DeepSeek-AI | Moonshot AI |
| 发布时间 | 2026-09-10 | 2026-07-16 |
| 模型类型 | 多模态大模型 | 推理大模型 |
| 架构 | MoE 架构 | MoE 架构 |
| 参数规模 | 5520亿 | 2.8万亿 |
| 上下文长度 | 1M | 1M |
| 最大输出 | 384K | 1M |
API 调用价格
价格优先使用 DataLearner 配置的 API 记录；缺失项不做推测。
| 价格项 | DeepSeek-V4.1-Flash | Kimi K3 |
|---|---|---|
| 文本输入 | ¥1 / 1M tokens | ¥20 / 1M tokens |
| 文本输出 | ¥4 / 1M tokens | ¥100 / 1M tokens |
| 缓存读取 | ¥0.02 / 1M tokens | ¥2 / 1M tokens |
小结
- DeepSeek-V4.1-Flash在以下类目领先:AI Agent - Tool Usage (4/4)、Coding and Software Engineer (3/3)、Multimodal Understanding (3/3)、Agent Level Benchmark (1/1)、General Knowledge (1/1)、Productivity Knowledge (1/1)
- Kimi K3在以下类目领先:科学与综合推理 (1/1)
- 胶着类目:Math and Reasoning
15 个共同 benchmark 上，DeepSeek-V4.1-Flash 平均高出 6.35 分。
单项差距最大的 benchmark：Terminal-Bench 4.0 — DeepSeek-V4.1-Flash 31.20，Kimi K3 12.60（分差 +18.60）。
本页正文由结构化模型、价格与 benchmark 数据生成，不使用实时 LLM 撰写。