Tag
#models
Every story tagged models, newest first.
Microsoft distilled a billion-parameter pathology model into 22M parameters
GigaPath-Flash keeps its teacher's accuracy to within 3 percent at roughly 50 times less compute, under Apache 2.0 with the weights on Hugging Face. The companion tumour model went the other way and beat its original, most of all out of distribution. Neither is validated for clinical use, and Microsoft says so first.
Ahmad J · Sep 2, 2026 · 4 min read
World Labs' Atlas wins six of seven 3D benchmarks, and you cannot run it
World Labs says its new Atlas world model beats five specialist reconstruction models. Averaged, it does. Row by row it wins six of seven, one of those by 0.2, and a one-billion-parameter open model takes the seventh. Access is limited to selected partners.
Ahmad J · Sep 2, 2026 · 5 min read
Claude Fable 5.1 costs 25% less, and its per-token price did not move
Anthropic's new model is cheaper for the reason nobody put in the headline: input and output rates are unchanged, and the whole discount is a 75 per cent cut to cache reads. That makes the saving conditional on how much context your workload re-reads, and the published benchmark gains are just as uneven.
Ahmad J · Sep 2, 2026 · 5 min read
DeepSeek's first V4 vision model is MIT-licensed, and 307 GB
DeepSeek shipped an open-weight multimodal model under MIT this weekend. Its own files carry three things the coverage left out: the download size, the fact that every benchmark is self-reported, and a footnote saying two of the wins are over a model that cannot see.
Ahmad J · Sep 1, 2026 · 5 min read
Small AI Models Are Quietly Winning in Production
Frontier models get the headlines, but inside real companies smaller, cheaper, faster models do the actual work. Here's how they win, where they don't, and what it costs to ignore them.
Ahmad J · Aug 14, 2026 · 4 min read
