OpenAI said on X on March 17, 2026 that GPT-5.4 mini was available in ChatGPT, Codex, and the API. The launch positions mini as a faster coding and multimodal workhorse, while OpenAI’s accompanying post also introduces GPT-5.4 nano for cheaper API-only workloads.
OpenAI introduced the Codex app on February 2, 2026. The macOS desktop interface is built to supervise multiple agents in parallel, manage skills and automations, and was expanded to Windows on March 4, 2026.
OpenAI Developers announced on March 20, 2026 that verified university students in the United States and Canada can claim $100 in Codex credits. OpenAI’s support page says that equals 2,500 ChatGPT credits, requires student verification through SheerID, and expires 12 months after the grant date.
OpenAI said on March 5, 2026 that GPT-5.4 Thinking and GPT-5.4 Pro were rolling out in ChatGPT, while GPT-5.4 also became available in the API and Codex. OpenAI’s launch page positions GPT-5.4 as a unified frontier model for reasoning, coding, native computer use, and long-horizon agent workflows.
OpenAI published a March 19, 2026 overview of its internal coding-agent monitoring stack. The company is using model-powered oversight in real deployments and argues similar safeguards should become standard for internal agent use.
OpenAI said on March 19, 2026 that it plans to acquire Astral. The deal would bring uv, Ruff, and ty into the Codex ecosystem and push Codex deeper into the full Python development workflow.
A front-page Hacker News discussion around Astral joining OpenAI captured both optimism about better AI-native Python tooling and concern about further consolidation of the developer stack.
A March 19, 2026 Hacker News post about Astral joining OpenAI reached 707 points and 445 comments at crawl time. Astral says it has agreed to join OpenAI as part of the Codex team while continuing to support Ruff, uv, and ty as open-source tools.
OpenAI said on X that GPT-5.4 mini is rolling out in ChatGPT, Codex, and the API, while GPT-5.4 nano is aimed at lower-cost API workloads. The company is positioning the pair as faster small models for coding, multimodal tasks, and agent sub-workflows.
OpenAI and Pacific Northwest National Laboratory introduced DraftNEPABench, a benchmark for permitting work that uses coding agents on document-heavy NEPA tasks. In tests across 18 federal agencies, experts found the system could save 1 to 5 hours per subsection, or roughly 15% of drafting time.
OpenAIDevs said on March 16, 2026 that subagents are now available in Codex. The feature lets developers keep the main context clean, split work across specialized agents, and steer individual threads as they run, while the official docs already describe PR review and CSV batch fan-out patterns.
OpenAI said on March 5, 2026 that GPT-5.4 is rolling out across ChatGPT, the API, and Codex. The new model combines GPT-5.3-Codex coding capability with OpenAI’s mainline reasoning stack, adds native computer-use features, and introduces experimental 1M-token context in Codex.