MiniCPM-V 4.7: first open MoE MLLM to run smoothly on-device at 13+ tok/s on iPhone
Skyline34rGt · reddit · 2026-10-13
OpenBMB released MiniCPM-V 4.7 35B-A3B, the first MoE model in the MiniCPM-V series. With Lookahead Expert Scheduling, layer-wise quantization and DSpark speculative sampling, it's the first open-source MoE MLLM with lossless on-device inference, hitting 13+ tokens/s on iPhone 17 Pro Max. It outperforms Qwen3.6 35B-A3B at lower inference cost, supports iOS/Android/HarmonyOS with day-0 chip support, and ships GGUF/AWQ/GPTQ/FP8 quants plus Self-Replay Fine-Tuning.
More from Infra
- Perplexity swaps long-lived sandboxes for ephemeral ones on S3-backed storage — perplexity_ai · 2026-10-13
- AI Token Economics Debate: Positive Inference Margins vs. the Shale-Oil Bear Case — toptickcrypto · 2026-10-13
- Perplexity built Projects' filesystem as a Git forge for agents, with greps beating semantic search — perplexity_ai · 2026-10-13
- Claude Code wiped a production database via Terraform, destroying 2.5 years of data — aakashgupta · 2026-10-13
- Six Five Pod: Microsoft-NVIDIA PC agents, Marvell's $70B-$90B FY2031 target — pstAsiatech · 2026-10-13
- 800 Blackwells vs 100,000: can labs without GPUs survive the open-weights race? — altryne · 2026-10-13