November 7, 2026
Ripple-65B Preview: open-weight model supports up to 1 million tokens of context
Ricursive's model can run on your own GPUs or be accessed via an API with per-token pricing.
Loading the page.
who it's about
about the same entities
Release
Ripple-65B Preview: an open model for AI agents with up to 1 million tokens of contextNovember 4
Insight
Agents write the code at 37signals, humans fix their bugs: DHH declares 'Pencils down'October 16
Insight
Cost per Task: one GLM-5.3 task costs 9x more at some providers than othersOctober 13
Release
Run the Pi agent on your own server: Pi pod keeps work sessions separateOctober 2
Insight
Same score, lower cost: Codex costs 3.2 times less than Claude Code in testsOctober 2
All news
One email a week - only what changes how you work. Subscribe