Nvidia earnings release - NVIDIA Newsroom

Ox Alpha (GLM 5.3 Flash) really moved the frontier. 3.8-Flash-Next is 1 point dumber, pricing unclear. Separately I'm surprised Qwen 27b is still so much more expensive than Gemma 4 26b.

Possible BS alert. You don’t need Fable to rewrite all the Cobol to Rust or Go, serve ads. OK Jane Street will pay a premium for the fastest, best models, but that might end-up being zero-sum sooner than expected. Do you really need $500b data centers to run all this stuff?

everybody’s an economist now

Follow the latest AI headlines via druce.ai on Bluesky