Hidden Codex Modal Hints at OpenAI's GPT-Image-2.5 Launch
Perplexity's own research-agent benchmark now crowns GPT-6 Astra, which scores 13.5% higher than Claude Fable 5.1 — while costing 6.1% less per task.
Codex
Most Recent 
OpenAI poaches Linear's product lead and Kairos co-founders for Codex and ChatGPT
Elon Musk says Grok 4.7 is arriving in 10 days with 2.1 trillion parameters and extra training on SpaceX data.
GLM Coding Plan turns one, and every subscriber gets a reset card
Elon Musk says Grok 4.7 is arriving in 10 days with 2.1 trillion parameters and extra training on SpaceX data.
a16z launches $1.1B 'Machine Age' fund for AI infrastructure
xAI has wired Grok Bot directly into X, giving paid users search, timeline reading, monitoring, and bookmark management — without needing to set up a developer account.
AI agents fail the real-world test: best score is just 20.6%
xAI has wired Grok Bot directly into X, giving paid users search, timeline reading, monitoring, and bookmark management — without needing to set up a developer account.
OpenAI's Codex is about to break another record, product lead hints
xAI has wired Grok Bot directly into X, giving paid users search, timeline reading, monitoring, and bookmark management — without needing to set up a developer account.
GLM-5.3 surges past GPT-5.6 Sol on Terminal-Bench 4.0
xAI has wired Grok Bot directly into X, giving paid users search, timeline reading, monitoring, and bookmark management — without needing to set up a developer account.
OpenAI fixes eight Codex quota-eating bugs, then resets your credits twice
xAI has wired Grok Bot directly into X, giving paid users search, timeline reading, monitoring, and bookmark management — without needing to set up a developer account.
Cognition's Devin hits $900M in annualized revenue — and may burn $800M this year
OpenAI is testing a persistent mode for Codex that would let the AI agent keep tracking and handling follow-up tasks after its main job is done. The code is already in the public repo, but there's no launch date.
OpenAI tests a Codex mode that keeps working after the task is done
OpenAI is testing a persistent mode for Codex that would let the AI agent keep tracking and handling follow-up tasks after its main job is done. The code is already in the public repo, but there's no launch date.