MiniCPM5 Trained Entirely on Huawei Ascend Tops Sub-2B Open-Weight Models
pstAsiatech · x · 2026-08-12
In May, ModelBest released MiniCPM5-1B, a 1-billion-parameter model. According to Artificial Analysis, it emerged as the top-performing open-weight model under 2 billion parameters, outperforming Alibaba's Qwen3.5-2B.
Notably, the model was trained entirely on Huawei's Ascend chips. ModelBest, which has kicked off its pre-IPO tutoring process, is leveraging its small-model strategy to optimize for domestic processors and bypass China's shortage of advanced AI chips.
More from Infra
- vLLM Teams Up with Microsoft and NVIDIA for Up to 7.3x Faster Model Loading on H100/A100 — vllm_project · 2026-08-12
- Skilled Labor Shortage Emerges as the Biggest Bottleneck for US AI Data Center Boom — coinfanking · 2026-08-12
- Bloom Energy Surges 14% as Data Centers Adopt On-Site Power for AI — BenBajarin · 2026-08-12
- Alibaba and Moonshot Target 10T Parameter Models, Hitting Memory Bottlenecks — pstAsiatech · 2026-08-12
- Base Power Raises $1B at $12B Valuation, Targeting AI Energy Dominance — Not Boring (Packy McCormick) · 2026-08-12
- Nvidia Partners with Wall Street for $500B AI Infrastructure Financing Deal — ocean_protocol · 2026-08-12