MiniCPM5-1B: A SOTA 1B on-device LLM, small yet powerful.
active 2024-12-21 → 2026-08-07 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 595 days from 2024-12-21 to 2026-08-07. Pushes: 114 total, peak 23 in a day. Pull requests: 17 total, peak 4 in a day. Issues: 79 total, peak 25 in a day. Comments: 71 total, peak 18 in a day. Stars: 1,015 total, peak 70 in a day.
- Pushes
- Pull requests
- Issues
- Comments
- Stars
Stars, PRs, issues and forks are under-captured in the later part of this window. GH Archive progressively stopped capturing non-push events during 2026 — −95% or worse by the end of the window. Every series here except Pushes fades for that reason, so a decline above reflects the archive, not this repository. Pushes stay reliable throughout, so read them, and the contributor counts derived from them, as the real signal. Data health has the measurements.
Top contributors
Pushes, PRs, issues, reviews and comments — stars and forks excluded, so this is contribution rather than popularity
| Contributor | Contributions | Pushes | PRs | Comments |
|---|---|---|---|---|
| xcjthu | 45 | 27 | 3 | 6 |
| zh-zheng | 45 | 1 | 1 | 18 |
| EEEEEKKO | 30 | 30 | 0 | 0 |
| BIGWangYuDong | 23 | 17 | 2 | 2 |
| Kaguya-19 | 14 | 7 | 2 | 3 |
| suhmily10 | 13 | 0 | 0 | 7 |
| Hytn | 12 | 11 | 0 | 1 |
| guanwy1995 | 12 | 3 | 1 | 4 |
| huangyuxiang03 | 6 | 5 | 0 | 1 |
| LDLINGLINGLING | 5 | 0 | 2 | 1 |
| jiahe7ay | 4 | 0 | 0 | 2 |
| saurabh12453 | 3 | 0 | 0 | 3 |
| XiaoduoAILab | 3 | 0 | 0 | 1 |
| agathonFong | 3 | 0 | 3 | 0 |
| lovangry | 3 | 3 | 0 | 0 |
| xf-hk | 3 | 0 | 0 | 1 |
| ZMXJJ | 3 | 2 | 0 | 1 |
| Weilun-Hub | 2 | 0 | 2 | 0 |
| xiaoganzi | 2 | 0 | 0 | 1 |
| avalonliberty | 2 | 0 | 0 | 0 |
Recent activity
Latest issues, pull requests and releases
- Pull request#353YOIMIYA662026-05-26 07:29
- Issue comment#352wangzhaode2026-05-26 05:28add MNN deployment cookbook and skill
- Issue#328guanwy19952026-02-12 07:57[Feature Request]: Details about minicpm-4 Eagle3 series models QAT process and quantization framework
- Issue comment#334BIGWangYuDong2026-02-12 03:19[Bad Case]: MiniCPM4-0.5B always outputs natural language instead of <tool_call> function calls
- Issue#325suhmily102026-02-12 03:08[Bad Case]: 使用minicpm4-0.5B和8B的模型都报错
- Issue#308suhmily102026-02-12 03:05Training instability with InfLLM v2 on MiniCPM4-8B using sparse attention
- Issue#312suhmily102026-02-12 03:04[Feature Request]: InfLLM v2特性是否与主流推理框架（如vLLM或SGLang）适配？
- Issue comment#312suhmily102026-02-12 03:04[Feature Request]: InfLLM v2特性是否与主流推理框架（如vLLM或SGLang）适配？
- Issue comment#327suhmily102026-02-12 03:01[Bad Case]: 微调MiniCPM 4.1-8B时，其loss显著高于其他模型
- Issue#326suhmily102026-02-12 02:58[Bad Case]: 使用minicpm4-0.5B同时启用sparse attention时报错
- Issue#330suhmily102026-02-12 02:57[Question] Model Parallel when Training MiniCPM4-8B
- Issue#329suhmily102026-02-12 02:55sparse attention torch实现
- Issue comment#331JFLFY22552026-02-11 12:49[Bad Case]: 关于jetson agx orin推理的问题
- Issue comment#324darkliang2025-12-18 11:52vllm video 如何设置fps？
- Issue#332xf-hk2025-12-14 07:49[Feature Request]: Has the training code for BitCPM been open-sourced? I haven't been able to find it.
- Issue#332xf-hk2025-12-14 07:49[Feature Request]: Has the training code for BitCPM been open-sourced? I haven't been able to find it.
- Issue comment#301xf-hk2025-12-14 07:43How to finetune BitCPM4-1B
- Issue#331alex2robotic2025-12-09 05:20[Bad Case]: 关于jetson agx orin推理的问题
- Issue comment#286dandenkijin2025-11-27 05:02[Question]: How can I find ArkInfer?
- Issue comment#324xcjthu2025-09-26 05:41vllm video 如何设置fps？
- Issue#324xcjthu2025-09-26 05:41vllm video 如何设置fps？
- Issue#325Yhc-7772025-09-25 02:09[Bad Case]: 使用minicpm4-0.5B和8B的模型都报错
- Issue comment#323suhmily102025-09-17 02:13slower than qwen 2.5 on a100 40gb
- Issue comment#323suhmily102025-09-15 12:57slower than qwen 2.5 on a100 40gb
- Pull request#322LDLINGLINGLING2025-09-12 11:29
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 1,015 stars here means stars gained during the window, not the repo's star count.