{"id":"mnn-pocket-0-3b","slug":"mnn-pocket-0-3b","name":"MNN-Pocket-0.3B","orgId":"taobao-mnn","family":"mnn-pocket-0-3b","releaseDate":null,"availability":"ga","openWeights":true,"license":"Apache-2.0 verified in model card and LICENSE.txt","paramsTotal":300000000,"paramsSource":"slug","contextWindow":null,"knowledgeCutoff":null,"modelCardUrl":"https://huggingface.co/taobao-mnn/MNN-Pocket-0.3B","hfRepo":null,"modalities":["text"],"aliases":["MNN-Pocket-0.3B","MNN Pocket 0.3B","taobao-mnn/MNN-Pocket-0.3B","MNN-Pocket-0.3B-GGUF","MNN-Pocket-0.3B-MNN"],"aliasNote":null,"origin":"tracked","tracked":{"trackedSince":"2026-11-08","announced":"2026-11-04","blurb":"The Alibaba Taobao MNN team's compact Chinese/English instruction-tuned text LLM, co-designed with its on-device MNN inference stack. Marketed as roughly 300M parameters (361.55M in HF safetensors metadata), with 32K deployed context and a lab-reported 173 MB INT4 W4A16 QAT MNN build. Apache-2.0 weights are public for local use; no public hosted API is confirmed.","api":"none","openWeights":"yes","params":{"text":"~300M advertised; 361.55M total safetensors parameters","source":"MNN team card and HF metadata"},"facts":[{"label":"Team and lineage","value":"MNN Team, Taobao and Tmall Group of Alibaba; instruction-tuned from taobao-mnn/MNN-Pocket-0.3B-Base. This is a separate MNN team release, not attributed to the Qwen lab merely because a qwen3 tag appears"},{"label":"Architecture","value":"28 layers, hidden size 896, GQA with 14 query / 2 KV heads, tied input/output embeddings, 65,536 vocabulary; BF16 native checkpoint"},{"label":"Context","value":"32K supported in deployment; configuration/card lists 40,960 positions, not a claim of 40K deployed support"},{"label":"License","value":"Apache-2.0 verified in model card and LICENSE.txt"},{"label":"Formats","value":"Public Transformers safetensors, GGUF and MNN formats. MNN INT4 W4A16 quantization-aware build is advertised at 173 MB; GGUF Q4_1 uses the QAT checkpoint"},{"label":"Limitations","value":"Team does not recommend reasoning-heavy advanced mathematics or long-horizon code generation; latency/throughput depends on hardware, backend and context"},{"label":"Related","value":"Separate from Base and intermediate SFT checkpoints, DFlash2 auxiliary variants, and the non-autoregressive Dec decision model. GGUF/MNN packaging variants stay on this instruction-model page"},{"label":"Date evidence","value":"Official alibaba/MNN README release note dated Nov 4, corroborated by a public HF discussion that day. Sep 23 repo creation and Sep 26 release-labelled commit do not independently prove historical public access"}],"claims":["Lab claim: INT4 W4A16 QAT deployment footprint is 173 MB. At 512-token context, decode reaches 466 tokens/s on iPad Pro M5 Metal (231 on its CPU); these are team measurements, not independent results.","Lab claim: average across the team's 11 Chinese/English benchmarks is 50.00 for native precision and 49.59 for INT4."],"sources":[{"kind":"repo","title":"MNN team model card and weights","url":"https://huggingface.co/taobao-mnn/MNN-Pocket-0.3B"},{"kind":"repo","title":"Official MNN README Nov 4 release news","url":"https://github.com/alibaba/MNN","date":"2026-11-04"},{"kind":"repo","title":"Public HF discussion corroborating Nov 4 access","url":"https://huggingface.co/taobao-mnn/MNN-Pocket-0.3B/discussions/1","date":"2026-11-04"},{"kind":"repo","title":"MNN-Pocket Apache-2.0 license","url":"https://huggingface.co/taobao-mnn/MNN-Pocket-0.3B/blob/main/LICENSE.txt"},{"kind":"lab","title":"MNN team technical report","url":"https://huggingface.co/taobao-mnn/MNN-Pocket-0.3B/blob/main/MNN-Pocket-0.3B-Technical-Report.pdf"}],"firstBenchmark":null},"organization":{"id":"taobao-mnn","name":"Alibaba Taobao MNN","slug":"taobao-mnn","country":null,"url":"https://github.com/alibaba/MNN","color":"#11110F","mark":"A"},"price":null,"answer":{"word":"Pending benchmarks.","scope":"Announced Nov 4, 2026; no independent results yet.","lead":"As of Nov 22, 2026, MNN-Pocket-0.3B is pending benchmarks on isitsota: announced Nov 4, 2026, with no independent results on the boards we track yet.","confidence":"low"},"status":"PENDING_BENCHMARKS","verdicts":[]}