Tag archive
Segment co-founder Calvin French-Owen argued that cheap fast models have crossed a usefulness threshold, pricing a personalized-news task he once ran for about a dollar at roughly ten cents, and the essay drew 499 points on Hacker News.
Where small open models genuinely beat frontier APIs after a fine-tune — classification, extraction, routing, on-device tasks — and where they still f
Liquid AI released LFM2.5-2.6B, a small model whose stack is mostly cheap convolution layers with attention rationed to eight of thirty, built for tool use and long context - and whose own model card says not to use it for agentic coding or knowledge
A developer fit a 28.9-million-parameter model onto an ESP32-S3 with 512KB of fast memory by keeping most of the model in flash storage and reading only about 450 bytes of it per word generated.
A new method called Program-as-Weights uses a 4-billion-parameter 'compiler' to turn a natural-language task description into a small weight file that a frozen 0.6B model runs, matching a 32B model's quality while using about one-fiftieth the memory
How Quillwright turns a photo and a voice note into a tradesperson's estimate, with an orchestra of...
Best GPUs for running Microsoft Phi-4 locally in 2026 — small 14B model that runs comfortably on $400 GPUs. VRAM and speed picks.
If 2025 was the year AI got a reality check 2026 is when the industry sobers up completely The shift...