OpenAI - Page 8

OpenAI says GPT-6 Astra and GPT-6.1 Sol score far higher on its own COBOL and Fortran coding evals, though production use still needs expert review.

Kapital agrees to buy YC-backed Woz for $30m cash and stock to deepen its AI banking stack. Closing not stated; patented token-reduction is Woz’s claim.

Emirates NBD and OpenAI say the bank will deploy GPT models on UAE in-country inference — a first-in-MENA claim. No value, model names or go-live date.

OpenCode Console users report unauthorised AI charges after a breach. Anthropic and Google blocked keys, OpenAI is refunding, and the incident is ongoing.

A newly unsealed filing by news publishers alleges Microsoft told staff to keep Project Mango, a crawler OpenAI paid it to run, secret. No court has ruled.

OpenAI changelog: prompt-to-voice GA for all API customers; nine new GPT-Live voices. GPT-Live only; max 100 voices/org. Docs only — no blog or X.

DeepSeek’s V4.1-Flash-1017 is a minor open-weight update. Its own table lifts many agent scores vs prior Flash; Opus 5 still leads HLE and Terminal-Bench.

OpenAI is rolling out a GPT-6 Pro ChatGPT upgrade for eligible Pro users: faster responses, better tool use, and multiple parallel agents.

Replit CEO Amjad Masad said Replit no longer works with Anthropic; an unnamed insider told TechCrunch the partnership is still in place. Disputed.

Goodfire’s Nick Cammarata argues neural nets are already mostly clean and classic superposition evidence was misread. Informal claim; critics push back.