Agentic ai now fits on a 2-billion-parameter edge model. minicpm5-2b, out september 9, brings tool…
agentic ai now fits on a 2-billion-parameter edge model. minicpm5-2b, out september 9, brings tool calling and multi-step reasoning to phones and iot devices without the cloud round-trip.
the small model is where the agent workload goes local.
Context
The MiniCPM5-2B model card on Hugging Face describes a dense Transformer built for on-device local deployment, with 2,516,756,480 parameters (1,981,982,720 non-embedding), a 131,072 context, a design for local assistants, coding agents and tool-use workflows, tool calling through an SGLang parser, and BFCL v4 66.6 within the card's own comparison set. It lists GGUF, MLX and GPTQ builds. In the same table Terminal-Bench v2.1 is 8.6 against 25.8 for the 4B-class model shown.
Tool calling is demonstrated on BFCL, a vendor-reported figure, while multi-step agent tasks are weak on the harder benchmark in the same card. Phones and IoT devices were not tested in the text read, and without the cloud round-trip follows from local running but was not measured. 2B is a size-class label, about 2.5 billion counting embeddings. Out September 9 is not on the card text read; an earlier note recorded a different date, so the release date is not established. The small model is where the agent workload goes local is the author's take.
Watch next
- On-device latency and memory measurements on real phones.
Sources
- MiniCPM5-2B model card (Hugging Face)huggingface.co
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 20 September 2026 at 23:19 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →