GLM-5.3 Flash12 minZ.ai's GLM-5.3-Flash is a 320B open model with 18B active parameters. One local tester crushed it to a 1-bit quant, maxed reasoning, and watched it ship a playable arcade game.Read now →
Model Wars
New model releases, benchmark fights and the race between the big labs. We explain what each launch changes in practice, not just what the press release claims.
6 storiesGLM-5.3 Flash12 minZ.ai's GLM-5.3-Flash is a 320B open model with 18B active parameters. One local tester crushed it to a 1-bit quant, maxed reasoning, and watched it ship a playable arcade game.Read now →
Claude course13 minBeginner Claude courses keep multiplying. The useful ones stop teaching button tours and start teaching model choice, Projects, artifacts, and when not to trust a fluent answer.Read now →
Week weird12 minSome weeks in AI are not one story — they are a pileup: architecture ideas like energy-based models, product drops, crypto-adjacent hacks, and IPO chatter. Here is how to hold the pileup without drowning.Read now →
Opus 5.512 minEvery few months a new Claude Opus drop makes builders say the same word: ridiculous. Under the hype is a more useful question — which jobs just moved from "try" to "trust with review"?Read now →
Weight surgery12 minTools and papers that let models inspect or edit weights sound like sci-fi sysadmin work. They point to a future where models are not only trained — they are maintained like living systems.Read now →
Claude ship12 minClaude product updates arrive fast enough to feel like a new app each month. Power users ignore the tour and rebuild a short list of workflows that print leverage.Read now →