OpenBMB Releases MiniCPM5-2B: A 2.52B Dense Model Averaging 53.9 Across 34 Benchmarks and Built to Run On Device

SummaryAI generated

OpenBMB released the MiniCPM5-2B large language model on September 7, 2026. This model has approximately 2.52 billion parameters and is designed with a dense architecture to achieve efficient on-side deployment. In the 34 benchmark tests it participated in, its average score was 53.9.

Related eventsRELATED EVENTS

2026-09-07 21:02The `GPT-6 Astra model was released, and OpenAI President Brockman confidently declared, “Welcome to the era of AGI.”
2026-09-08 19:21OpenAI released GPT-6 Astra, marking the arrival of the AGI era, with capabilities for autonomous computer operation and scientific research.
2026-09-08 22:40① Nvidia CEO Jensen Huang posted that OpenAI’s GPT-6 Astra, released last week, was trained using approximately 100,000 NV Link 72 clusters, and he believes that General Artificial Intelligence (AGI) has officially arrived; ② GPT-6 Astra can directly operate computers and software to perform complex tasks such as programming, reaching the most advanced level in multiple fields. OpenAI has announced the beginning of the AGI era.

Key entitiesKEY ENTITIES

LlamaForCausalLMMiniCPM5-2BOpenBMBQwen3.5-4Bgranite-4.2-3B

Event frameEVENT FRAME

Launch

OpenBMB MiniCPM5-2B 开源2.52B参数大模型，34项基准平均分53.9

Coverage · reports per dayLANGUAGE SPLIT

Entity relations

SignalsSIGNALS

Keyword heat

OpenBMB1
MiniCPM5-2B1
Qwen3.5-4B1
granite-4.2-3B1
LlamaForCausalLM1

All reports (1)SOURCES

↗

M MarkTechPost en 2026-09-08 03:18

OpenBMB 发布了 MiniCPM5-2B，这是一个拥有 2.52B 参数的稠密语言模型。该模型采用标准 Llama 架构，原生上下文窗口为 131,072 tokens，支持 vLLM、SGLang 等主流引擎部署。在 34 个基准测试中平均得分为 53.9，优于同尺寸类中的 Qwen3.5-4B（51.1）。模型在工具使用、代码推理及长上下文检索任务上表现突出，但在通用知识方面落后于大模型。训练流程包含 SFT、RL 及基于 JustRL II 算法的 on-policy 蒸馏，并开源了 UltraData 系列数据集及中间检查点。