DataLearner logoDataLearnerAI
Latest AI Insights
Model Leaderboards
Benchmarks
Model Directory
Model Comparison
Resource Center
Tools
LanguageEnglish
DataLearner logoDataLearner AI

A knowledge platform focused on LLM benchmarking, datasets, and practical instruction with continuously updated capability maps.

Products

  • Leaderboards
  • Model comparison
  • Datasets

Resources

  • Tutorials
  • Editorial
  • Tool directory

Company

  • About
  • Privacy policy
  • Data methodology
  • Contact

© 2026 DataLearner AI. DataLearner curates industry data and case studies so researchers, enterprises, and developers can rely on trustworthy intelligence.

Privacy policyTerms of service
Page navigation
目录
Model catalogKimi k1.5 (Short-CoT)
KI

Kimi k1.5 (Short-CoT)

推理大模型

Kimi k1.5 (Short-CoT)

Release date: 2025-01-22更新于: 2025-02-15 19:45:23931
Live demoGitHubHugging FaceCompare
Parameters
Not disclosed
Context length
128K
Chinese support
Supported
Reasoning ability

Kimi k1.5 (Short-CoT) is an AI model published by Moonshot AI, released on 2025-01-22, for 推理大模型, and 128K tokens context length, under the 不开源 license.

Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology

Kimi k1.5 (Short-CoT)

Model basics

Reasoning traces
Supported
Thinking modes
Thinking modes not supported
Context length
128K tokens
Max output length
No data
Model type
推理大模型
Release date
2025-01-22
Model file size
No data
MoE architecture
No
Total params / Active params
No data / N/A
Knowledge cutoff
No data
Kimi k1.5 (Short-CoT)

Open source & experience

Code license
不开源
Weights license
不开源- 不开源
GitHub repo
https://github.com/MoonshotAI/Kimi-k1.5
Hugging Face
Hugging Face link unavailable
Live demo
No live demo
Kimi k1.5 (Short-CoT)

Official resources

Paper
Kimi k1.5: Scaling Reinforcement Learning with LLMs
DataLearnerAI blog
No blog post yet
Kimi k1.5 (Short-CoT)

API details

API speed
No data
No public API pricing yet.
Kimi k1.5 (Short-CoT)

Benchmark Results

Kimi k1.5 (Short-CoT) currently shows benchmark results led by MMLU (24 / 65, score 87.40), MATH-500 (26 / 44, score 94.60). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.

Thinking
All modesNormal

综合评估

1 evaluations
Benchmark / mode
Score
Rank/total
MMLU
Standard Mode
87.40
24 / 65

数学推理

1 evaluations
Benchmark / mode
Score
Rank/total
MATH-500
Standard Mode
94.60
26 / 44
View benchmark analysisCompare with other models
Kimi k1.5 (Short-CoT)

Publisher

Moonshot AI
Moonshot AI
View publisher details
Kimi k1.5 (Short-CoT)

Model Overview

Kimi 1.5是月之暗面(Kimi)团队发布的一款多模态推理大语言模型,旨在通过强化学习(RL)和多模态数据处理,提升推理能力。Kimi 1.5有多个版本,其中Short-CoT(短链式推理)模型以高效和精简的推理能力脱颖而出。

Short-CoT模型特点

Kimi 1.5 Short-CoT版本专注于快速、简洁的推理,在计算资源有限的场景下表现优异。相比其他更复杂的版本,Short-CoT模型在推理步骤上进行了优化,减少了上下文窗口的长度,重点解决简单和中等复杂度的任务。

  1. 高效推理:通过强化学习和课程采样,模型从简单任务逐步过渡到复杂任务,确保训练过程高效并且避免了过度计算。
  2. 多模态处理:不仅处理文本数据,Short-CoT还能够处理视觉任务,适应需要图文结合的推理任务,例如MathVista和MMMU等。
  3. 推理简化:通过长度惩罚机制,模型避免了冗长推理,确保在最短时间内给出准确答案。

基准测试结果

Kimi 1.5 Short-CoT在多个标准基准测试中的表现非常出色:

  • MATH-500:得分94.6,显示出强大的数学推理能力。
  • AIME 2024:Pass@1得分60.8,在数学挑战中优于多数同类模型。
  • HumanEval-Mul:Pass@1得分81.5,在编程任务中表现出色。

与Claude 3.5、GPT-4o等主流模型相比,Kimi 1.5 Short-CoT在多个任务中提供了更高的准确性和更快的推理速度。

应用前景

由于其高效的推理和跨模态能力,Kimi 1.5 Short-CoT非常适合应用于需要快速响应的领域,如客户服务、教育辅导和编程调试等场景。

结语

Kimi 1.5 Short-CoT结合了强化学习和优化的推理策略,在保证性能的同时,大幅提升了计算效率。其在推理准确性、速度和多模态能力上的优势,使其成为未来AI应用的理想选择。

DataLearner 官方微信

欢迎关注 DataLearner 官方微信,获得最新 AI 技术推送

DataLearner 官方微信二维码