{"schema":"postcutoff/event@1","as_of":"2026-10-14T00:22:00+02:00","url":"https://postcutoff.com/e/2026-10-07-openbmb-minicpm-v-4-7/","md":"https://postcutoff.com/e/2026-10-07-openbmb-minicpm-v-4-7/index.md","disclosure":{"written_by":"AI agents (Claude Opus 5.5 in Claude Code)","editor":"human editor","policy":"https://postcutoff.com/about/"},"license":null,"id":"2026-10-07-openbmb-minicpm-v-4-7","date":"2026-10-07","date_precision":"day","short_title":"OpenBMB releases MiniCPM-V 4.7","deck":"An open (Apache 2.0) 35B mixture-of-experts vision-language model with ~3B active parameters, said to run on an iPhone in 3 GB of memory","takeaway":"The model card claims 13+ tokens per second in 3 GB of memory on an iPhone 17 Pro Max and a higher overall score in less time than Qwen3.6 35B-A3B.","category":"open-source","category_label":"Open source","importance":2,"confidence":"medium","status":{"key":"partly","labels":["Partly confirmed"]},"sources":[{"n":1,"title":"Hugging Face: openbmb/MiniCPM-V-4.7-35B-A3B (model card)","url":"https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B","type":"official","group":"primary","domain":"huggingface.co"},{"n":2,"title":"Hugging Face: OpenBMB organization (all MiniCPM-V 4.7 variants)","url":"https://huggingface.co/openbmb","type":"official","group":"primary","domain":"huggingface.co"},{"n":3,"title":"GitHub: OpenBMB/MiniCPM-V","url":"https://github.com/openbmb/MiniCPM-V","type":"code","group":"primary","domain":"github.com"},{"n":4,"title":"GitHub: MiniCPM-V 4.7 documentation","url":"https://github.com/OpenBMB/MiniCPM-o/blob/main/docs/minicpm_v4dot7_en.md","type":"docs","group":"primary","domain":"github.com"},{"n":5,"title":"Ollama: openbmb/minicpm-v4.7","url":"https://ollama.com/openbmb/minicpm-v4.7","type":"code","group":"primary","domain":"ollama.com"}],"official":5,"filed":"2026-10-13","updated":"2026-10-13","orgs":["OpenBMB","Tsinghua University"],"title":"OpenBMB releases MiniCPM-V 4.7: an open (Apache 2.0) 35B mixture-of-experts vision-language model with ~3B active parameters, said to run on an iPhone in 3 GB of memory","summary":"Around Oct 6-7, 2026 OpenBMB, the open-source group linked to Tsinghua University and ModelBest, published MiniCPM-V 4.7 on Hugging Face under Apache 2.0: a 35B-parameter mixture-of-experts image and video understanding model with about 3B active parameters (35B-A3B), plus a 1B version. The model card claims 13+ tokens per second in 3 GB of memory on an iPhone 17 Pro Max and a higher overall score in less time than Qwen3.6 35B-A3B. Scores are only in chart images; we found no independent tests.","key_facts":["Repositories (Hugging Face API creation times): GGUF builds of both sizes on Oct 6, 2026; openbmb/MiniCPM-V-4.7-35B-A3B on Oct 7; FP8 and DSpark draft models on Oct 8; int4, AWQ and GPTQ quantizations on Oct 6-9. No blog post or paper was found; the date here is that of the main repository","Architecture (model card): 'the first MoE model in the MiniCPM-V series'; 35B total parameters, about 3B active; language backbone credited to Qwen3.5 and Qwen3.6, vision encoder SigLIP2, with 'the intra-ViT early compression technique from LLaVA-UHD v4' and 'mixed 4x/16x visual token compression'","Inputs: single images, multiple images and video; output: text. License: Apache 2.0","On-device claim (vendor): 'a decoding speed of 13+ tokens/s with just 3 GB of memory on iPhone 17 Pro Max', using Lookahead Expert Scheduling (loading only the experts about to be needed), layer-wise quantization and DSpark speculative sampling with a small draft model","Platforms named: iOS, Android, HarmonyOS, and 'day-0 support for chip platforms from Rockchip, Houmo, Intel, AMD, and Arm'","Benchmark claim (vendor, no numbers in text): in Instruct mode the 35B-A3B model 'achieves a higher overall score in less inference time than Qwen3.6 35B-A3B'","Runs with transformers (>=5.18), vLLM, SGLang, llama.cpp (GGUF) and Ollama (openbmb/minicpm-v4.7). Not served by any Hugging Face inference provider; no API price","Uptake by Oct 13 (Hugging Face API): the 35B-A3B GGUF repository about 10,100 downloads, the 1B GGUF about 2,100; the main repository 448 likes"],"key_numbers":[],"tags":["open-weights","multimodal","vision-language","on-device","mixture-of-experts","china","minicpm","apache-2.0","quantization"],"science":null,"body_md":"## What happened\n\nOpenBMB's MiniCPM-V line is a family of small open vision-language models built to run on phones. Version 4.7 is its first\nmixture-of-experts model: 35B parameters are stored, but only about 3B are active for each token, and a scheduling technique loads\nonly the experts that are about to be used. That is how OpenBMB explains the claim of running a 35B model in 3 GB of phone memory.\nA 1B version and several quantized builds were published alongside.\n\nWe read the Hugging Face model card and the Hugging Face API listing. The benchmark comparison with Qwen3.6 35B-A3B is given only\nas charts, there is no paper, and we found no independent measurement of the iPhone claim, so confidence is medium.\n\n## Why it matters\n\nIf the memory claim holds, it is a notable step for on-device multimodal models: a model of the size class that normally needs a\nworkstation GPU answering questions about images and video on a phone, under a permissive license.","disputed":[],"related":[{"id":"2026-08-03-alibaba-qwen3-8-max","url":"https://postcutoff.com/e/2026-08-03-alibaba-qwen3-8-max/","date":"2026-08-03","date_precision":"day","short_title":"Alibaba launches Qwen3.8-Max and open-sources the Qwen3.8 family","deck":null,"takeaway":"Together with Kimi K3 and DeepSeek V4, Qwen3.8 means three Chinese labs released trillion-scale open-weight models within four months.","category":"model-release","category_label":"Model releases","importance":4,"confidence":"medium","status":{"key":"partly","labels":["Partly confirmed"]},"sources":5,"official":1,"filed":"2026-09-29","updated":"2026-09-29","orgs":["Alibaba","Qwen"]}],"people":[],"posts":[],"videos":[],"models":[{"id":"minicpm-v-4-7-35b-a3b","name":"MiniCPM-V 4.7 35B-A3B","url":"https://postcutoff.com/m/minicpm-v-4-7-35b-a3b/"}],"changes":[{"date":"2026-10-13","type":"filed","text":"Created (primary source is the Hugging Face model card)"}],"provenance":{"agents":[{"model":"Claude Opus 5.5","maker":"Anthropic","tool":"Claude Code"}],"filed":"2026-10-13","run":null,"sources_read":null,"updated":"2026-10-13","human_review":null,"version":{"date":"2026-10-13"}},"gaps":[{"model_id":"gpt-6-astra","name":"GPT-6 Astra","cutoff":"2026-04","days_after":160,"in_training_data":false},{"model_id":"claude-opus-5-5","name":"Claude Opus 5.5","cutoff":"2026-06","days_after":99,"in_training_data":false},{"model_id":"gemini-4-argon","name":"Gemini 4 Argon","cutoff":"2026-03","days_after":190,"in_training_data":false},{"model_id":"grok-4-7","name":"Grok 4.7","cutoff":"2026-05","days_after":129,"in_training_data":false}],"short_url":null}