OpenBMB open-sources MiniCPM-V 4.7: 35B multimodal models running on phones
Aiden_Tech_Ai · x · 2026-10-13
OpenBMB released the open-source MiniCPM-V 4.7 multimodal LLM series, pushing the performance-efficiency Pareto frontier. The 35B-A3B model brings SOTA-level performance to local devices with a new Lookahead Expert Scheduling (LES) technique that predicts and preloads experts, while a 1B variant targets lightweight deployment. The team reports competitive results across 34 benchmarks spanning multimodal understanding, STEM reasoning, OCR and long-video comprehension, with efficient on-device inference on iOS and Android.
Related event: OpenBMB Open-Sources MiniCPM-V 4.7: 35B MoE Multimodal On-Device(13 posts)→
More from Infra
- AMD Ramps Helios Rack Into Production a Year After Debut, Unveils Helion Rack and Verano CPU — ryanshrout · 2026-10-13
- Can you mix an RTX 5060 Ti with a 7900XTX for LLM work? CUDA + ROCm together — hedgehog0 · 2026-10-13
- n+2 lookahead expert prefetching for Qwen Flash Next SSD streaming nets +8% on Mac — turtleninja99 · 2026-10-13
- Muse Wonder requests die with 502 exactly 300s after last output on OpenRouter — adonis_singh · 2026-10-13
- SonicBoom: A Python-Free C++ Runtime Beats llama.cpp Decode by 14% on a Laptop 3050 — nalaginrut · 2026-10-13
- GP builds local clinical assistant on one RTX 4090: Gemma 4 26B QAT cuts response to ~3 seconds — looneycheetah · 2026-10-13