Codex Ships Decisions GA, Uber Rollout

// 1h agoPRODUCT UPDATE

Codex Ships Decisions GA, Uber Rollout

OpenAI’s Day 18 Codex/Work update pairs the Decisions API’s general availability with performance improvements, Open Responses joining the Linux Foundation’s Agentic AI Foundation, and Uber rolling ChatGPT Work out to more than 30,000 employees while engineers continue using Codex. The post also promises a usage reset before 10 a.m. PT.

// ANALYSIS

This is a platform-consolidation day disguised as a daily Codex ship: OpenAI is improving the execution layer, standardizing the surrounding ecosystem, and proving enterprise demand.

–Decisions API exits beta with shared-input caching, structured JSON input, image support, and up to 400 questions per request.
–Open Responses under neutral Linux Foundation governance could accelerate multi-provider compatibility while reducing vendor lock-in.
–Uber’s 30,000-plus employee rollout is a meaningful enterprise validation for ChatGPT Work, with Codex retained for software engineering.
–The unspecified performance fixes may matter most in long-running agent workflows, but the post provides no metrics to evaluate them.
–The reset keeps the 28-day campaign engaging, while also blending product news with subscription-retention messaging.

// TAGS

codexchatgpt-workai-codingcoding-agentagentapiopen-sourcedevtool

DISCOVERED

1h ago

2026-10-23

PUBLISHED

2h ago

2026-10-23

RELEVANCE

9/ 10

AUTHOR

thsottiaux

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

RESEARCH28m ago

System One Models Challenge LLM Legal Certainty

A new arXiv study tests four System One decision models against six generative LLMs on legal-authority verification, including 1,670 cases where the correct holding is missing. The best decision model matched the top LLM’s 88.1% standard-task accuracy at roughly 1/55 the cost, while combining both approaches reduced wrong citations.

OPEN SOURCE37m ago

MiniMax M3.1 Flash Opens 428B Weights

MiniMax has released M3.1 Flash’s open weights on Hugging Face and ModelScope. The mixture-of-experts model has 428B total parameters, 23B activated per token, and targets agentic coding and tool-use workloads.

MODEL49m ago

OlmoEarth v1.3 opens planet-scale satellite embeddings

Ai2 released OlmoEarth v1.3, an open satellite foundation model that compresses annual Sentinel-1, Sentinel-2, and Landsat imagery into 128-dimensional embeddings for each 10-meter land pixel. Precomputed global data and a no-login browser explorer make crop mapping, land-cover analysis, and change detection far more accessible.