jeinlee1991/chinese-llm-benchmark

Skip(New discovery — enrichment pending)

非线智能 NoneLinear - ReLE评测：中文AI大模型能力评测（持续更新）：目前已囊括374个大模型，覆盖chatgpt、gpt-5.4、谷歌gemini-3.1-pro、Claude-4.6、文心ERNIE-X1.1、ERNIE-5.0、qwen3.6-max、qwen3.6-plus、百川、讯飞星火、商汤senseChat等商用模型，以及step3.5-flash、kimi-k2.6、ernie4.5、MiniMax-M2.7、deepseek-v4、Qwen3.6、llama4、智谱GLM-5.1、MiMo-V2、LongCat、gemma4、mistral等开源大模型。不仅提供排行榜，也提供规模超200万的大模型缺陷库！方便广大社区研究分析、改进大模型。

Why this repo matters

Benchmarks 374 Chinese and global LLMs including latest commercial and open-source models for comprehensive capability comparison.
Provides a two-million-entry defect database enabling community research into model failures and improvement opportunities.
Covers agentic AI evaluation topics relevant to developers building reliable agent systems with Chinese language support.

Key Metrics

Stars: 6.4k
Forks: 263
Open Issues: 18
Stars (7d delta): 0
Stars (30d delta): 0

Scores & Metadata

Trend Velocity Score: 0.00
Opportunity Score: 10.00
Confidence Score: 65.00
Topics:
🏷️ agentic-ai 🏷️ artificial-intelligence 🏷️ llm-agent 🏷️ llm-evaluation
First Seen: 2 months ago

Star History

6.4k stars+20 over 3 snapshots8/27/2026 – 9/5/2026

Links & Sources

Related Repositories

obra/superpowers

An agentic skills framework & software development methodology that works.

292.4kShell

mattpocock/skills

Skills for Real Engineers. Straight from my .agents directory.

272.6kShell

forrestchang/andrej-karpathy-skills

A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls.

212.1k

DigitalPlatDev/FreeDomain

DigitalPlat FreeDomain: Free Domain For Everyone

200kMarkdown