Kimi K2.6vsGLM 5.1
在 18 个共同 benchmark 中，Kimi K2.6 整体领先：Kimi K2.6 领先 14 项，GLM 5.1 领先 4 项，持平 0 项，平均分差 +10.73。
Kimi K2.614 项(78%)(22%)4 项GLM 5.1
评测分数
按能力类目分组，每组内按分差大小排列；共 18 项。
Agent Level Benchmark
Kimi K2.6 领先 2/3| 评测项 | Kimi K2.6 | GLM 5.1 | 分差 |
|---|---|---|---|
| τ³-Banking | 23.3082 / 164Thinking (With Tools) | 13.60115 / 164Thinking (With Tools) | +9.70 |
| τ²-Bench - Telecom | 93.9043 / 264Normal (With Tools) | 97.1017 / 264Normal (With Tools) | -3.20 |
| Terminal Bench Hard | 37.9059 / 244Normal (With Tools) | 35.6070 / 244Normal (With Tools) | +2.30 |
AI Agent - Tool Usage
Kimi K2.6 领先 3/3| 评测项 | Kimi K2.6 | GLM 5.1 | 分差 |
|---|---|---|---|
| Tool Decathlon | 502 / 10Thinking (With Tools) | 40.706 / 10Thinking (With Tools) | +9.30 |
| Terminal-Bench 2.1 | 65.9094 / 192Thinking (With Tools) | 61.80105 / 192Thinking (With Tools) | +4.10 |
| Terminal Bench 2.0 | 66.7010 / 48Thinking (With Tools) | 63.5013 / 48Thinking (With Tools) | +3.20 |
Coding and Software Engineer
Kimi K2.6 领先 2/2| 评测项 | Kimi K2.6 | GLM 5.1 | 分差 |
|---|---|---|---|
| SciCode | 51.5061 / 130Thinking (No Tools) | 44.8090 / 130Thinking (No Tools) | +6.70 |
| SWE-Bench Pro - Public | 58.6018 / 62Thinking (With Tools) | 58.4020 / 62Thinking (With Tools) | +0.20 |
Math and Reasoning
Kimi K2.6 领先 2/2| 评测项 | Kimi K2.6 | GLM 5.1 | 分差 |
|---|---|---|---|
| IMO-AnswerBench | 8610 / 24Thinking (No Tools) | 83.8014 / 24Thinking (No Tools) | +2.20 |
| AIME 2026 | 96.409 / 29Thinking (No Tools) | 95.3012 / 29Thinking (No Tools) | +1.10 |
AI Agent - Information Search
Kimi K2.6 领先 1/1| 评测项 | Kimi K2.6 | GLM 5.1 | 分差 |
|---|---|---|---|
| BrowseComp | 83.2016 / 57Thinking (With Tools + Internet) | 79.3019 / 57Thinking (With Tools + Internet) | +3.90 |
General Knowledge
GLM 5.1 领先 1/1| 评测项 | Kimi K2.6 | GLM 5.1 | 分差 |
|---|---|---|---|
| HLE | 19.60290 / 563Normal (No Tools) · Text only | 27.90226 / 563Normal (No Tools) · Text only | -8.30 |
Instruction Following
GLM 5.1 领先 1/1| 评测项 | Kimi K2.6 | GLM 5.1 | 分差 |
|---|---|---|---|
| IF Bench | 44.30184 / 282Normal (No Tools) | 52148 / 282Normal (No Tools) | -7.70 |
Long Context
Kimi K2.6 领先 1/1| 评测项 | Kimi K2.6 | GLM 5.1 | 分差 |
|---|---|---|---|
| AA-LCR | 69.70106 / 170Normal (No Tools) | 53.30134 / 170Normal (No Tools) | +16.40 |
Multimodal Understanding
Kimi K2.6 领先 1/1| 评测项 | Kimi K2.6 | GLM 5.1 | 分差 |
|---|---|---|---|
| GDP.pdf | 1365 / 118Thinking (No Tools) | 8.4086 / 118Thinking (No Tools) | +4.60 |
Text Embedding
Kimi K2.6 领先 1/1| 评测项 | Kimi K2.6 | GLM 5.1 | 分差 |
|---|---|---|---|
| Context Arena | 51.8886 / 126Normal (No Tools) | 30.29116 / 126Normal (No Tools) | +21.59 |
Writing and Creative Capabilities
Kimi K2.6 领先 1/1| 评测项 | Kimi K2.6 | GLM 5.1 | 分差 |
|---|---|---|---|
| Creative Writing | 1,72125 / 106Normal (No Tools) | 1,58940 / 106Normal (No Tools) | +132.10 |
科学与综合推理
GLM 5.1 领先 1/1| 评测项 | Kimi K2.6 | GLM 5.1 | 分差 |
|---|---|---|---|
| GPQA Diamond | 78.80249 / 462Normal (No Tools) | 83.90180 / 462Normal (No Tools) | -5.10 |
规格对比
| 字段 | Kimi K2.6 | GLM 5.1 |
|---|---|---|
| 发布机构 | Moonshot AI | 智谱AI |
| 发布时间 | 2026-04-20 | 2026-03-27 |
| 模型类型 | 推理大模型 | 推理大模型 |
| 架构 | MoE 架构 | MoE 架构 |
| 参数规模 | 1万亿 | 7540亿 |
| 上下文长度 | 256K | 200K |
| 最大输出 | 暂无数据 | 125K |
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 | Kimi K2.6 | GLM 5.1 |
|---|---|---|
| 文本输入 | $0.95 / 1M tokens | $1.4 / 1M tokens |
| 文本输出 | $4 / 1M tokens | $4.4 / 1M tokens |
| 缓存读取 | $0.16 / 1M tokens | $4.4 / 1M tokens |
| 缓存写入 | $0.95 / 1M tokens | $0.26 / 1M tokens |
小结
- Kimi K2.6在以下类目领先:Agent Level Benchmark (2/3)、AI Agent - Tool Usage (3/3)、Coding and Software Engineer (2/2)、Math and Reasoning (2/2)、AI Agent - Information Search (1/1)、Long Context (1/1)、Multimodal Understanding (1/1)、Text Embedding (1/1)、Writing and Creative Capabilities (1/1)
- GLM 5.1在以下类目领先:General Knowledge (1/1)、Instruction Following (1/1)、科学与综合推理 (1/1)
18 个共同 benchmark 上，Kimi K2.6 平均高出 10.73 分。
单项差距最大的 benchmark：Creative Writing — Kimi K2.6 1,721，GLM 5.1 1,589（分差 +132.10）。
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。