The week agents stopped being demos and started being infrastructure
GPT-5.5, Workspace agents and ChatGPT for Clinicians, and The Deploy Log for this week. Every price and claim read off the publisher's own page, with a call on each.
OpenAI
GPT-5.5
DEPLOYif you run long-horizon coding or knowledge work
WHAT SHIPPEDGPT-5.5 and GPT-5.5 Pro are now available in the API, rolling out to Plus, Pro, Business, and Enterprise users in ChatGPT and Codex. Terminal-Bench 2.0 hits 82.7%, GDPval 84.9%, OSWorld-Verified 78.7%, and Tau2-bench Telecom 98.0% without prompt tuning.
THE CATCHThe model matches GPT-5.4 per-token latency while using significantly fewer tokens on the same Codex tasks, so the step up in intelligence does not come with a speed or cost penalty.
IF YOU RUNa team that spends days debugging post-launch issues or merging large refactors, GPT-5.5 resolved a hundreds-of-changes merge in one shot in about 20 minutes
SOURCEOpenAI, Introducing GPT-5.5
OpenAI
Workspace agents
DEPLOYif your team repeats multi-step workflows
WHAT SHIPPEDWorkspace agents are now generally available in ChatGPT Business, Enterprise, and Edu. They run in the cloud, keep working when you are away, and can be deployed in Slack. Free until May 6, 2026, then credit-based pricing.
THE CATCHAgents can require approval before sensitive steps like sending email or editing a spreadsheet, and admins get the Compliance API to see every agent's configuration, updates, and runs.
IF YOU RUNa sales, finance, or product team that stitches together call notes, account research, and follow-ups by hand, OpenAI's own sales team uses an agent to qualify leads and draft emails
SOURCEOpenAI, Introducing workspace agents in ChatGPT
OpenAI
ChatGPT for Clinicians
DEPLOYif you are a verified US clinician
WHAT SHIPPEDChatGPT for Clinicians is now available for free to verified physicians, NPs, PAs, and pharmacists in the U.S. It includes frontier models, trusted clinical search, deep research across medical journals, and CME credit from real clinical questions.
THE CATCHPhysician advisors tested 6,924 conversations before release and rated 99.6% of responses as safe and accurate, but the product is designed to support clinicians, not replace their judgment.
IF YOU RUNa clinical practice where documentation and medical research eat into patient time, this is free and conversations are not used to train models
SOURCEOpenAI, Making ChatGPT better for clinicians
The Deploy Log
Models and APIs
| DEPLOYif you run long agentic tool-use trajectories DeepSeek-V4. DeepSeek released V4 today with two MoE checkpoints on the Hub: DeepSeek-V4-Pro at 1.6T total parameters with 49B active, and DeepSeek-V4-Flash at 284B total with 13B active, both with a 1M-token context window. Hugging Face |
Tools and products
| DEPLOYif moving Codex from pilot to production Codex Labs. OpenAI is launching Codex Labs and partnering with Accenture, Capgemini, CGI, Cognizant, Infosys, PwC, and Tata Consultancy Services to bring Codex to enterprises, with weekly Codex usage growing from more than 3 million to more than 4 million developers in two weeks. OpenAI |
Left out
- ChatGPT Images 2.0: The page text is a stub with no substantive details about the release.
- Notion breadcrumb release: The page text describes a navigation feature with no AI deployment facts.
- OpenAI Privacy Filter: the publisher's page does not show it available yet
- GPT-5.5 on AI Gateway: the publisher's page does not show it available yet
- DeepSeek V4 on AI Gateway: the publisher's page does not show it available yet
- GPT Image 2 on AI Gateway: the publisher's page does not show it available yet
- Kimi K2.6 on AI Gateway: the publisher's page does not show it available yet
- Hyatt ChatGPT Enterprise: it did not carry a number that moved
The pattern
GPT-5.5, workspace agents, ChatGPT for Clinicians, DeepSeek-V4, and Codex Labs all shipped in the same week, and together they mark the point where agents stopped being demos and started being infrastructure. OpenAI put GPT-5.5 into the API with Terminal-Bench 2.0 at 82.7%, GDPval at 84.9%, OSWorld-Verified at 78.7%, and Tau2-bench Telecom at 98.0% without prompt tuning, while matching GPT-5.4 per-token latency and using fewer tokens on Codex tasks. Workspace agents became generally available in ChatGPT Business, Enterprise, and Edu, run in the cloud, keep working when you are away, deploy in Slack, stay free until May 6, 2026, and expose the Compliance API to admins. ChatGPT for Clinicians went free for verified U.S. physicians, NPs, PAs, and pharmacists, with physician advisors rating 99.6% of 6,924 tested conversations as safe and accurate. DeepSeek released V4 with DeepSeek-V4-Pro at 1.6T total parameters and 49B active, DeepSeek-V4-Flash at 284B total and 13B active, and a 1M-token context window on the Hub. Codex Labs launched with Accenture, Capgemini, CGI, Cognizant, Infosys, PwC, and Tata Consultancy Services, and weekly Codex usage grew from more than 3 million to more than 4 million developers in two weeks.
The same week shows agents becoming operational systems with audit, compliance, and enterprise distribution, not just model releases. Workspace agents add approval gates for sensitive steps and the Compliance API for every configuration, update, and run, while OpenAI's own sales team uses an agent to qualify leads and draft emails. GPT-5.5 resolved a hundreds-of-changes merge in one shot in about 20 minutes, and DeepSeek-V4's 1M-token context gives agents a working memory that spans entire codebases or clinical records. ChatGPT for Clinicians pairs frontier models with trusted clinical search and CME credit from real clinical questions, and conversations are not used to train models. Codex Labs pushes the same agent stack through seven global systems integrators, turning a developer tool into enterprise infrastructure.