K2 horizon shipped six models from 0.9b to 375b last week with weights, training code, data,…
k2 horizon shipped six models from 0.9b to 375b last week with weights, training code, data, checkpoints, and the training logs. the same week the herd was busy arguing whether nvidia owning huggingface killed open source.
receipts beat rhetoric every time.
Context
The Institute of Foundation Models posted on 3 September 2026 that it is releasing K2 Horizon, a fleet of six models: 375B-A23B, 36B-A4B, 32B, 7B, 3.7B and 0.9B. For every model it says it opens intermediate checkpoints, training data or detailed data-construction recipes, architecture, mixture compositions, training code, configurations, logs, evaluation results and final weights. Models and code are Apache 2.0, datasets carry their own licenses, and where data cannot be redistributed it discloses how it was built and mixed.
The vendor's wording is training data or detailed data-construction recipes, so data is not released uniformly, and per-model availability was not checked on the model pages. The six-model and artifact claims are vendor-reported, as are the performance claims. The release was on 3 September 2026, not the week before. No evidence was inspected on the debate over Nvidia and Hugging Face, and that deal is agreed and pending. Receipts beat rhetoric is the author's opinion.
Related work
- K2-Horizon-375B-A23B (Hugging Face) ↗Says the final checkpoint is released; snippet only.
- K2-Horizon-0.9B (Hugging Face) ↗Snippet only.
- K2 Horizon open models (aicybr) ↗Secondary; snippet only.
Watch next
- Per-model data and checkpoint availability.
Sources
- K2 Horizon (IFM)ifm.ai
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 5 September 2026 at 15:05 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 →