# Week ahead: OpenAI DevDay with no public agenda, and two Gemini models switch off

> OpenAI's keynote is Tuesday, Micron reports Wednesday, and two Gemini API model IDs stop working on September 30 and October 2.

Source: Promptea — https://www.promptea.me/en/blog/week-ahead-2026-09-27-openai-devday-gemini-shutdowns-micron · Language: English · Last updated: 2026-09-27

Published: 2026-09-27 · Event date: 2026-09-27 · Category: other · Author: Promptea Editorial · Covers: 2026-09-27 – 2026-10-03

**Summary:** The confirmed AI calendar for September 27 to October 3 is short: OpenAI DevDay on September 29, a Gemini preview shutdown and Micron's earnings call on September 30, CoreWeave's Fully Connected conference, and a gemini-2.5-flash-image shutdown on October 2. Claude Sonnet 4.5's September 29 date is a floor, not a retirement.

This is a short week on the calendar, and that is the honest version of it. Only a handful of AI-relevant dates are publicly confirmed for September 27 to October 3. One of them is a developer conference whose content nobody outside OpenAI has seen; two are hard shutdown dates in the Gemini API that will break code if you ignore them. Everything below has a date stated by the organizer. Nothing here is a prediction.

## Tuesday, September 29: OpenAI DevDay

OpenAI holds [DevDay 2026](https://devday.openai.com/) at Fort Mason in San Francisco. The opening keynote, featuring Sam Altman, starts at 10:00 a.m. Pacific (14:00 in Buenos Aires) and will be livestreamed free; OpenAI says the other sessions will be recorded and posted afterwards. The in-person event was application-only, and applications [closed in July](https://community.openai.com/t/openai-devday-2026-applications-are-now-open/1384509). OpenAI also says DevDay Exchanges will follow later in 2026 in eight cities, including São Paulo and Mexico City, without giving dates on the event page.

What OpenAI has **not** published is an agenda of announcements. The event page promises technical sessions on APIs and tools and demos, nothing more specific. Any list of expected launches you see this weekend is guesswork. For context only: OpenAI's [API changelog](https://developers.openai.com/api/docs/changelog) shows it shipped GPT-6 Sol and GPT-6 Luna on September 22 and put the Agents API into public beta on September 10, so the platform already moved a lot this month before the keynote.

**For developers:** if you build on the OpenAI API, the useful move is to watch the keynote and then read the changelog, not the recap threads. Pricing, rate limits and deprecations tend to show up in the documentation, not on stage.

## Wednesday, September 30: a Gemini preview model shuts down

Google's [Gemini API deprecation table](https://ai.google.dev/gemini-api/docs/deprecations) lists **gemini-omni-flash-preview** with a shutdown date of September 30, 2026. The listed replacement is **gemini-omni-1.1-flash**, released August 27. After the shutdown date, calls to the retired model ID will stop working, so any hard-coded reference to the preview model needs to move before Wednesday.

## Wednesday, September 30: Micron reports its fiscal fourth quarter

Micron [scheduled its earnings call](https://www.globenewswire.com/news-release/2026/08/26/3351673/14450/en/micron-technology-to-report-fiscal-fourth-quarter-results-on-september-30-2026.html) for September 30 at 2:30 p.m. Mountain time (17:30 in Buenos Aires), announced by press release on August 26. Micron is a major supplier of high-bandwidth memory for AI accelerators, so its results and outlook are one of the few hard, company-reported data points on AI hardware demand this week. We will not guess the numbers; the call is where the company's own figures land.

## September 29 to October 1: CoreWeave Fully Connected

CoreWeave's Fully Connected conference runs in San Francisco from September 29 to October 1. NVIDIA's [event page](https://www.nvidia.com/en-us/events/coreweave-fully-connected/) says Ian Buck, NVIDIA's VP of Hyperscale and High-Performance Computing, joins the Day 1 keynote on September 30 to talk about AI factory infrastructure, and that the keynote can be watched live. This is an infrastructure-customer event, not a product launch venue we can vouch for in advance.

## Friday, October 2: gemini-2.5-flash-image shuts down

The same Google table lists **gemini-2.5-flash-image** with a shutdown date of October 2, 2026, exactly a year after its release. The recommended replacement is **gemini-3.1-flash-image-preview**. Note the detail: Google is pointing users of a stable image model to a **preview** model. Preview models can change and usually carry their own, shorter shutdown schedules, so teams running image generation in production should test the replacement and check its terms rather than just swapping the ID.

## A date that is not a deadline: Claude Sonnet 4.5

Anthropic's [model deprecations page](https://platform.claude.com/docs/en/about-claude/model-deprecations) lists claude-sonnet-4-5-20250929 with a tentative retirement date of "not sooner than September 29, 2026." That is a floor, not a schedule. The model is still marked Active, no deprecation has been announced, and Anthropic says it gives at least 60 days' notice before retiring a publicly released model. So Sonnet 4.5 cannot be switched off this week. If you still use it, the floor passing is a reason to plan a migration, not to rush one. The same page lists Claude Haiku 4.5 with a floor of October 15, 2026, also without a deprecation notice.

## What did not make this list

We also looked for AI-relevant earnings calls and state-law effective dates in this window. Where we could not open a primary source stating the date, we left the item out rather than repeat a secondhand calendar.

## Key takeaways

- Sep 29: OpenAI DevDay keynote at 10:00 a.m. PT (14:00 Buenos Aires), livestreamed free.
- Sep 30: gemini-omni-flash-preview shuts down; move to gemini-omni-1.1-flash. Micron's earnings call is at 2:30 p.m. MT.
- Oct 2: gemini-2.5-flash-image shuts down; Google's recommended replacement is a preview model.
- Claude Sonnet 4.5 is not being retired this week: its date is a floor and Anthropic promises 60 days' notice.

## Why this matters

- Two Gemini API model IDs have hard shutdown dates this week; code still calling them will fail after September 30 and October 2.
- OpenAI's DevDay keynote is the week's most likely source of API changes, but OpenAI has published no agenda, so any launch list is speculation.
- Micron's results are one of the few hard, company-reported data points on AI memory demand this week.

## Sources

- [OpenAI DevDay 2026](https://devday.openai.com/) — OpenAI (primary source), accessed 2026-09-27
- [OpenAI DevDay 2026 applications are now open!](https://community.openai.com/t/openai-devday-2026-applications-are-now-open/1384509) — OpenAI Developer Community (primary source), accessed 2026-09-27
- [API changelog](https://developers.openai.com/api/docs/changelog) — OpenAI (primary source), accessed 2026-09-27
- [Gemini API: Gemini deprecations](https://ai.google.dev/gemini-api/docs/deprecations) — Google AI for Developers (primary source), accessed 2026-09-27
- [Micron Technology to Report Fiscal Fourth Quarter Results on September 30, 2026](https://www.globenewswire.com/news-release/2026/08/26/3351673/14450/en/micron-technology-to-report-fiscal-fourth-quarter-results-on-september-30-2026.html) — Micron Technology (GlobeNewswire) (primary source), accessed 2026-09-27
- [NVIDIA at CoreWeave Fully Connected](https://www.nvidia.com/en-us/events/coreweave-fully-connected/) — NVIDIA (primary source), accessed 2026-09-27
- [Model deprecations](https://platform.claude.com/docs/en/about-claude/model-deprecations) — Anthropic (Claude Platform docs) (primary source), accessed 2026-09-27

*Researched with AI assistance and verified against primary sources before publishing. Every claim links to where it came from.*
