MiniCPM Models by OpenBMB
Last refreshed 2026-10-22. Next refresh: weekly.
Details
Capabilities
About
MiniCPM is a family of compact yet powerful language models developed by OpenBMB, designed for efficient deployment on edge devices including smartphones, PCs, and automotive systems. Despite their small size (typically 1-8B parameters), MiniCPM models achieve performance competitive with much larger models. All models are released as open-weight on HuggingFace.
Decision facts
- Best fit
- vision and multimodal workreasoningJSON / Tool use
- Capability starting point
- MiniCPM-V 4.7 1B with 256K context and reasoning, JSON / Tool use, and multimodal inputs
- Lowest tracked input
- Not tracked
- Closest related family
- 360Zhinao
Current Variants
Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.
Use when the workload needs vision, 256K context, and 1.6B parameters.
Use when the workload needs vision, 256K context, and reasoning.
Use when the workload needs 128K context, 2.5B parameters, and reasoning.
Use when the workload needs 1.1B parameters, reasoning, and JSON / Tool use.
Use when the workload needs 256K context, 1.3B parameters, and JSON / Tool use.
Use when the workload needs 32K context and 8B parameters.
Use when the workload needs 4K context and 2.4B parameters.
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| MiniCPM-V 4.7 1B | Use when the workload needs vision, 256K context, and 1.6B parameters. | 2026-10 | vision256K context1.6B parameters | Current |
| MiniCPM-V 4.7 35B-A3B | Use when the workload needs vision, 256K context, and reasoning. | 2026-10 | vision256K contextreasoning | Current |
| MiniCPM5-2B | Use when the workload needs 128K context, 2.5B parameters, and reasoning. | 2026-09 | 128K context2.5B parametersreasoning | Current |
| MiniCPM5-1B | Use when the workload needs 1.1B parameters, reasoning, and JSON / Tool use. | 2026-05 | 1.1B parametersreasoningJSON / Tool use | Current |
| MiniCPM-V 4.6 | Use when the workload needs 256K context, 1.3B parameters, and JSON / Tool use. | 2026-05 | 256K context1.3B parametersJSON / Tool use | Current |
| MiniCPM-4 8B | Use when the workload needs 32K context and 8B parameters. | 2025-05 | 32K context8B parameters | Current |
| MiniCPM 2B | Use when the workload needs 4K context and 2.4B parameters. | 2024-02 | 4K context2.4B parameters | Current |
Release Timeline
5 release groupsSpecifications(7 models)
| Model | Released | Context | Parameters | Vision | Multimodal | Reasoning | JSON / Tool use |
|---|---|---|---|---|---|---|---|
| MiniCPM-V 4.7 1B | 2026-10 | 256K | 1.55B | Yes | Yes | Yes | Yes |
| MiniCPM-V 4.7 35B-A3B | 2026-10 | 256K | 35.2B total, ~3B active | Yes | Yes | Yes | Yes |
| MiniCPM5-2B | 2026-09 | 128K | 2.5B | No | No | Yes | Yes |
| MiniCPM5-1B | 2026-05 | — | 1.1B | No | No | Yes | Yes |
| MiniCPM-V 4.6 | 2026-05 | 256K | 1.3B | Yes | Yes | No | Yes |
| MiniCPM-4 8B | 2025-05 | 32K | 8B | No | No | No | No |
| MiniCPM 2B | 2024-02 | 4K | 2.4B | No | No | No | No |
Popular comparisons in this family
- MiniCPM-V 4.6 vs Qwen3.5-9B272
- MiniCPM-V 4.6 vs Qwen2-7B-Instruct76
- MiniCPM-V 4.6 vs Mistral Large 2 (2407)41
- MiniCPM-V 4.6 vs Mistral Small 426
- MiniCPM 2B vs Nemotron Mini 4B Instruct21
- MiniCPM 2B vs Qwen3.5-9B18
- MiniCPM 2B vs Qwen2-7B-Instruct16
- Together AI - Gemma 3n-e4B vs MiniCPM 2B14
- MiniCPM-V 4.6 vs Mistral Large 3 675B Instruct13
- MiniCPM 2B vs Mistral 7B v0.111
Models(7)
MiniCPM-V 4.7 1B
MiniCPM-V 4.7 35B-A3B
MiniCPM5-2B
MiniCPM5-1B
MiniCPM-V 4.6
MiniCPM-4 8B
MiniCPM 2B