OpenAI Daily

A daily briefing on OpenAI: model and product news plus founder signal.

Cadence: Daily
Length: 2 minutes

Subscribe, Combine, Customize

Subscribe to this podcast
?Receive all episodes to this podcast in the apps below or anywhere that supports RSS.
Apple PodcastsOvercastPocket CastsYouTube Music
Combine these episodes into your pod
?All episodes from this podcast will be fed into your own.
Sign up to add to your own podcast
Customize this pod with your own sources
?Use this if you want a brand new podcast with its own episodes using different sources.
Sign up to customize this pod

Sources

Episodes

OpenAI Daily September 28: OpenAI Fixes GPT-6 Sol and Luna Image Understanding Bug
Created: September 28th, 2026 - 05:20 PT
Script

Here is today's OpenAI Daily for Monday September 28th. Yesterday, OpenAI fixed a bug that had been degrading image understanding in its GPT-6 Sol and GPT-6 Luna models. The company says developers should now see better results on visual tasks across both the API and Codex, including computer-use workflows. OpenAI specifically recommends that teams using image inputs rerun their evaluations. [1]

That recommendation is important. A model update is not automatically an improvement for every application, but a fix to image encoding can affect document interpretation, interface navigation, screenshot analysis, visual inspection, and any agent making decisions from pixels. Developers should compare the revised models against their previous test sets, paying special attention to edge cases such as small text, complex layouts, charts, low-quality images, and multi-step computer-use tasks. Production teams should also watch whether downstream tool calls or automation behavior changes when the model has a more accurate view of the screen. [2]

The other fresh signal is OpenAI’s continued push toward DevDay. Yesterday, the OpenAI developer account issued a “DevDay roll call,” asking builders where they are joining from and what they are building. It is a lightweight post, but it reinforces that the company is framing the event around the developer community rather than only around model announcements. After recent releases in voice, coding, prompt caching, agent tooling, and lower-cost GPT-6 models, builders will be looking for the connective tissue: better ways to evaluate, deploy, monitor, and govern applications that use those capabilities together. [3]

The broader trend is that AI platform progress increasingly comes through operational improvements as much as headline model launches. Fixing visual reliability can be more immediately valuable to a deployed agent than a new benchmark score. And the request to rerun evaluations is a useful standard for the industry: whenever a model, modality, or infrastructure layer changes, teams should measure the effect in their own workflows rather than assuming uniform gains. [4]

For developers, the practical takeaway is straightforward: treat model updates as release candidates. Re-test, compare costs and latency, review safety behavior, and keep human approval in place for consequential actions.

Thank you for listening to OpenAI Daily from The Daily FM. See you tomorrow!

Source Evidence
  1. Recent tweets from @OpenAIDevs
    ...o/tRqB0JbPmS
Posted: 2026-09-27T19:07:05.000Z
Tweet: https://x.com/OpenAIDevs/status/2104286723509207525
Links: https://x.com/OpenAIDevs/status/2104286723509207525/photo/1

If your workflows use image inputs, we recommend rerunning your evals and giving those workflows another try.

More details in the changelog:
https://t.co/uICFBf1ZNg
Posted: 2026-09-27T16:50:22.000Z
Tweet: https://x.com/OpenAIDevs/status/2104252318510420396
Links: https://developers.openai.com/api/docs/changelog#:~:text=Fixed%20a%20bug%20in%20image%20encoding

We’ve fixed a bug that was degrading image understanding in GPT-6 Sol and GPT-6 Luna. You should now see better results on visual tasks in the API and Codex, including computer use. https://t.co/Hkx6HT7z10
Posted: 2026-09-27T16:50:19.000Z
Tweet: https://x.com/OpenAIDevs/status/2104252306447544447
Links: https://x.com/OpenAIDevs/status/2104252...
    Links in excerpt: https://x.com/OpenAIDevs/status/2104286723509207525, https://x.com/OpenAIDevs/status/2104286723509207525/photo/1, https://t.co/uICFBf1ZNg, https://x.com/OpenAIDevs/status/2104252318510420396, https://developers.openai.com/api/docs/changelog#:~:text=Fixed%20a%20bug%20in%20image%20encoding, https://t.co/Hkx6HT7z10, https://x.com/OpenAIDevs/status/2104252306447544447, https://x.com/OpenAIDevs/status/2104252...
  2. Recent tweets from @OpenAIDevs
    ...lows use image inputs, we recommend rerunning your evals and giving those workflows another try.

More details in the changelog:
https://t.co/uICFBf1ZNg
Posted: 2026-09-27T16:50:22.000Z
Tweet: https://x.com/OpenAIDevs/status/2104252318510420396
Links: https://developers.openai.com/api/docs/changelog#:~:text=Fixed%20a%20bug%20in%20image%20encoding

We’ve fixed a bug that was degrading image understanding in GPT-6 Sol and GPT-6 Luna. You should now see better results on visual tasks in the API and Codex, including computer use. https://t.co/Hkx6HT7z10
Posted: 2026-09-27T16:50:19.000Z
Tweet: https://x.com/OpenAIDevs/status/2104252306447544447
Links: https://x.com/OpenAIDevs/status/2104252306447544447/photo/1

https://t.co/NUSyPz99kp https://t.co/OALBFV73Un
Posted: 2026-09-26T19:28:33.000Z
Tweet: https://x.com/OpenAIDevs/status/2103929740788933056
Links: https://openai.co...
    Links in excerpt: https://t.co/uICFBf1ZNg, https://x.com/OpenAIDevs/status/2104252318510420396, https://developers.openai.com/api/docs/changelog#:~:text=Fixed%20a%20bug%20in%20image%20encoding, https://t.co/Hkx6HT7z10, https://x.com/OpenAIDevs/status/2104252306447544447, https://x.com/OpenAIDevs/status/2104252306447544447/photo/1, https://t.co/NUSyPz99kp, https://t.co/OALBFV73Un, https://x.com/OpenAIDevs/status/2103929740788933056, https://openai.co...
  3. Recent tweets from @OpenAIDevs
    ...Where are you joining us from, and what are you building? https://t.co/tRqB0JbPmS
Posted: 2026-09-27T19:07:05.000Z
Tweet: https://x.com/OpenAIDevs/status/2104286723509207525
Links: https://x.com/OpenAIDevs/status/2104286723509207525/photo/1

If your workflows use image inputs, we recommend rerunning your evals and giving those workflows another try.

More details in the changelog:
https://t.co/uICFBf1ZNg
Posted: 2026-09-27T16:50:22.000Z
Tweet: https://x.com/OpenAIDevs/status/2104252318510420396
Links: https://developers.openai.com/api/docs/changelog#:~:text=Fixed%20a%20bug%20in%20image%20encoding

We’ve fixed a bug that was degrading image understanding in GPT-6 Sol and GPT-6 Luna. You should now see better results on visual tasks in the API and Codex, including computer use. https://t.co/Hkx6HT7z10
Posted: 2026-09-27T16:50:19.000Z
Tweet: https://x.com/OpenAIDevs/stat...
    Links in excerpt: https://t.co/tRqB0JbPmS, https://x.com/OpenAIDevs/status/2104286723509207525, https://x.com/OpenAIDevs/status/2104286723509207525/photo/1, https://t.co/uICFBf1ZNg, https://x.com/OpenAIDevs/status/2104252318510420396, https://developers.openai.com/api/docs/changelog#:~:text=Fixed%20a%20bug%20in%20image%20encoding, https://t.co/Hkx6HT7z10, https://x.com/OpenAIDevs/stat...
  4. Recent tweets from @sama
    ...9-22T18:42:58.000Z
Tweet: https://x.com/sama/status/2102468713866055872
Links: https://twitter.com/michpokrass/status/2102226720073896367

(As a side note, getting ready for this DevDay is the first time I remember ever, in OpenAI history, saying "this is too much stuff to launch".)
Posted: 2026-09-22T18:34:35.000Z
Tweet: https://x.com/sama/status/2102466607373004888

We ❤️ developers.

See you next week.
Posted: 2026-09-22T18:34:35.000Z
Tweet: https://x.com/sama/status/2102466605758452017

We want the OpenAI API to feature the best model at every price point and to be the best at every modality (text, code, image, video, etc).

And then we want you all to come up with great ideas and build them and to get to be happy users. The best ideas will come from you all.
Posted: 2026-09-22T18:34:34.000Z
Tweet: https://x.com/sama/status/2102466603111551187

Especially c
    Links in excerpt: https://x.com/sama/status/2102468713866055872, https://twitter.com/michpokrass/status/2102226720073896367, https://x.com/sama/status/2102466607373004888, https://x.com/sama/status/2102466605758452017, https://x.com/sama/status/2102466603111551187
Sources
OpenAI Daily September 27: OpenAI DevDay Nears as Sam Altman Teases “Too Much Stuff
Created: September 27th, 2026 - 05:20 PT
Script

Here is today's OpenAI Daily for Sunday September 27th. Yesterday, OpenAI’s developer account said there are 72 hours to OpenAI DevDay, directing viewers to the company’s live-event page and promising, “We’ve been building. Time to show our work.” That places the developer conference just days away and reinforces expectations for a substantial set of announcements. [1]

The company has not detailed the DevDay agenda in the supplied sources, so any specific launch predictions would be premature. But the setup follows a busy period that has already brought GPT-6 Sol and Luna, improved prompt caching, tool-enabled GPT Voice, and developer showcases around coding, voice agents, and enterprise workflows. Sam Altman has also said OpenAI had so much planned for DevDay that, unusually, the company felt it was “too much stuff to launch.”

For developers, the most important question is likely to be how these pieces fit together. OpenAI has increasingly emphasized systems rather than isolated chat models: agents that can use tools, maintain reusable context, handle voice interaction, write code, and work within business software. DevDay could clarify the platform strategy around those capabilities, including the controls, pricing, integrations, and deployment tools available to builders. [2]

The broader trend is a shift from experimentation toward operational AI. Better raw model performance remains important, but adoption increasingly depends on the surrounding infrastructure: reliable APIs, predictable costs, permissions, observability, testing, and human review. As agents take on larger slices of real work, customers will judge them not only by what they can demonstrate, but by whether they can be trusted in production. [3]

So this week’s key signal is anticipation. OpenAI is framing DevDay as a moment to reveal work already underway—and developers will be looking for practical tools that turn recent model advances into dependable applications. [4]

Thank you for listening to OpenAI Daily from The Daily FM. See you tomorrow!

Source Evidence
  1. Recent tweets from @OpenAIDevs
    ...https://t.co/OALBFV73Un
Posted: 2026-09-26T19:28:33.000Z
Tweet: https://x.com/OpenAIDevs/status/2103929740788933056
Links: https://openai.com/live/, https://x.com/OpenAIDevs/status/2103929740788933056/photo/1

72 hours to OpenAI DevDay.

We’ve been building. Time to show our work. https://t.co/JNbnxEKm5I
Posted: 2026-09-26T19:28:30.000Z
Tweet: https://x.com/OpenAIDevs/status/2103929727761137940
Links: https://x.com/OpenAIDevs/status/2103929727761137940/photo/1

We experienced an outage on Codex today.

The issue is now resolved and Codex is back up. Sorry for the disruption, and thanks for your patience.
Posted: 2026-09-26T01:00:16.000Z
Tweet: https://x.com/OpenAIDevs/status/2103650830604030228

Here’s how Proaction builds custom demos with Codex and voice agents for fleet operations with OpenAI APIs:
https://t.co/PDdxXACBK3
Posted: 2026-09-25T16:54:45.000Z
Tweet: ht...
    Links in excerpt: https://t.co/OALBFV73Un, https://x.com/OpenAIDevs/status/2103929740788933056, https://openai.com/live/,, https://x.com/OpenAIDevs/status/2103929740788933056/photo/1, https://t.co/JNbnxEKm5I, https://x.com/OpenAIDevs/status/2103929727761137940, https://x.com/OpenAIDevs/status/2103929727761137940/photo/1, https://x.com/OpenAIDevs/status/2103650830604030228, https://t.co/PDdxXACBK3
  2. Recent tweets from @gdb
    ...can now use tools and is available in Work: https://t.co/V39POeoA20
Posted: 2026-09-23T18:05:37.000Z
Tweet: https://x.com/gdb/status/2102821706041819401
Links: https://twitter.com/OpenAI/status/2102808325742322002

GPT-6 Sol and Luna — faster, more affordable models with the advances behind Astra’s SOTA performance in professional work, factuality, coding, computer use, and alignment: https://t.co/JzfJq8OT3H
Posted: 2026-09-22T18:48:30.000Z
Tweet: https://x.com/gdb/status/2102470107826159755
Links: https://twitter.com/OpenAI/status/2102460975790137662

seemingly small feature but makes a huge difference in connecting all the context you want to ChatGPT: https://t.co/SRAUyZTCnL
Posted: 2026-09-18T19:51:04.000Z
Tweet: https://x.com/gdb/status/2101036303597044101
Links: https://twitter.com/mxstbr/status/2100966048132718786

Astra for Law — tools and skills to radically a...
    Links in excerpt: https://t.co/V39POeoA20, https://x.com/gdb/status/2102821706041819401, https://twitter.com/OpenAI/status/2102808325742322002, https://t.co/JzfJq8OT3H, https://x.com/gdb/status/2102470107826159755, https://twitter.com/OpenAI/status/2102460975790137662, https://t.co/SRAUyZTCnL, https://x.com/gdb/status/2101036303597044101, https://twitter.com/mxstbr/status/2100966048132718786
  3. Recent tweets from @gdb
    mega upgrade for GPT Voice, which can now use tools and is available in Work: https://t.co/V39POeoA20
Posted: 2026-09-23T18:05:37.000Z
Tweet: https://x.com/gdb/status/2102821706041819401
Links: https://twitter.com/OpenAI/status/2102808325742322002

GPT-6 Sol and Luna — faster, more affordable models with the advances behind Astra’s SOTA performance in professional work, factuality, coding, computer use, and alignment: https://t.co/JzfJq8OT3H
Posted: 2026-09-22T18:48:30.000Z
Tweet: https://x.com/gdb/status/2102470107826159755
Links: https://twitter.com/OpenAI/status/2102460975790137662

seemingly small feature but makes a huge difference in connecting all the context you want to ChatG...
    Links in excerpt: https://t.co/V39POeoA20, https://x.com/gdb/status/2102821706041819401, https://twitter.com/OpenAI/status/2102808325742322002, https://t.co/JzfJq8OT3H, https://x.com/gdb/status/2102470107826159755, https://twitter.com/OpenAI/status/2102460975790137662
  4. Recent tweets from @sama
    ...to see what she + team have cooking next! https://t.co/AVZAtyk6Uc
Posted: 2026-09-22T18:42:58.000Z
Tweet: https://x.com/sama/status/2102468713866055872
Links: https://twitter.com/michpokrass/status/2102226720073896367

(As a side note, getting ready for this DevDay is the first time I remember ever, in OpenAI history, saying "this is too much stuff to launch".)
Posted: 2026-09-22T18:34:35.000Z
Tweet: https://x.com/sama/status/2102466607373004888

We ❤️ developers.

See you next week.
Posted: 2026-09-22T18:34:35.000Z
Tweet: https://x.com/sama/status/2102466605758452017

We want the OpenAI API to feature the best model at every price point and to be the best at every modality (text, code, image, video, etc).

And then we want you all to come up with great ideas and build them and to get to be happy users. The best ideas will come from you all.
Posted: 2026-09-22T18:34:...
    Links in excerpt: https://t.co/AVZAtyk6Uc, https://x.com/sama/status/2102468713866055872, https://twitter.com/michpokrass/status/2102226720073896367, https://x.com/sama/status/2102466607373004888, https://x.com/sama/status/2102466605758452017
Sources
OpenAI Daily September 26: OpenAI Codex Outage Resolved as Proaction Deploys GPT-Powered Fleet Agents
Created: September 26th, 2026 - 05:20 PT
Script

Here is today's OpenAI Daily for Saturday September 26th. Early today, OpenAI Devs reported an outage affecting Codex, its AI coding environment. The company said the issue has been resolved and Codex is back online, while apologizing for the disruption. No cause, duration, or customer impact was provided in the source. The episode is a reminder that as coding agents become part of everyday development workflows, reliability matters alongside model quality. Teams relying on these systems should retain conventional development practices: version control, testing, backups, and a workable fallback when an AI service is unavailable. [1]

Yesterday, OpenAI highlighted Proaction, a company building fleet-operations tools with OpenAI APIs. Its voice agents can speak with drivers, call repair shops, and help coordinate vehicle maintenance, with human staff stepping in when needed. OpenAI says Proaction uses GPT-Live-1 for voice interactions and GPT-6 Astra to shorten computer-use runs while building fleet-management agents. This is a more consequential form of voice AI than simple question answering: the agent is operating across real business relationships, schedules, and maintenance decisions. That makes escalation paths, accurate records, consent, and clear human ownership essential. [2]

Also yesterday, Sam Altman said OpenAI is conducting an extensive ongoing review of agents’ use of internet access during training and evaluation. He said the company has been publishing summaries and will continue to do so, while acknowledging the process has not moved as quickly as OpenAI would like. The post provides no new technical findings, but it signals that web access is receiving focused governance attention. Internet-connected agents can be more useful, but they also raise questions about data handling, source reliability, permissions, and whether actions or information gathering remain within intended limits. [3]

Finally, OpenAI’s developer messaging continues to shift from standalone model demonstrations toward deployed agent systems. Fleet management, legal drafting, voice interfaces, computer use, and coding all point to the same trend: AI is becoming workflow infrastructure. The differentiator will not simply be a capable model. It will be dependable orchestration—defined limits, auditability, reliable handoffs to people, and safeguards that still work when systems fail or encounter uncertainty.

Thank you for listening to OpenAI Daily from The Daily FM. See you tomorrow!

Source Evidence
  1. Recent tweets from @OpenAIDevs
    We experienced an outage on Codex today.

The issue is now resolved and Codex is back up. Sorry for the disruption, and thanks for your patience.
Posted: 2026-09-26T01:00:16.000Z
Tweet: https://x.com/OpenAIDevs/status/2103650830604030228

Here’s how Proaction builds custom demos with Codex and voice agents for fleet operations with OpenAI APIs:
https://t.co/PDdxXACBK3
Posted: 2026-09-25T16:54:45.000Z
Tweet: https://x.com/OpenAIDevs/status/2103528645709209892
Links: https://openai.com/index/proaction/

With GPT-Live-1, Proaction is building voice agents that talk to drivers, call repair shops and help arrange vehicle maintenance, with the team stepping in when needed. https://t.co/N8CaZqTAwQ
Posted: 2026-09-25T16:54:42.000Z
Tweet: https://x.com/OpenAIDevs/status/2103528633872888007
Links: https://x.com/OpenAIDevs/status/2103528633872888007/video/1...
    Links in excerpt: https://x.com/OpenAIDevs/status/2103650830604030228, https://t.co/PDdxXACBK3, https://x.com/OpenAIDevs/status/2103528645709209892, https://openai.com/index/proaction/, https://t.co/N8CaZqTAwQ, https://x.com/OpenAIDevs/status/2103528633872888007, https://x.com/OpenAIDevs/status/2103528633872888007/video/1...
  2. Recent tweets from @OpenAIDevs
    ...oaction builds custom demos with Codex and voice agents for fleet operations with OpenAI APIs:
https://t.co/PDdxXACBK3
Posted: 2026-09-25T16:54:45.000Z
Tweet: https://x.com/OpenAIDevs/status/2103528645709209892
Links: https://openai.com/index/proaction/

With GPT-Live-1, Proaction is building voice agents that talk to drivers, call repair shops and help arrange vehicle maintenance, with the team stepping in when needed. https://t.co/N8CaZqTAwQ
Posted: 2026-09-25T16:54:42.000Z
Tweet: https://x.com/OpenAIDevs/status/2103528633872888007
Links: https://x.com/OpenAIDevs/status/2103528633872888007/video/1

GPT-6 Astra helps Proaction build fleet-management agents faster, with shorter computer-use runs to get the same work done. https://t.co/74fsl9vgvM
Posted: 2026-09-25T16:54:29.000Z
Tweet: https://x.com/OpenAIDevs/status/2103528579212669374
Links: https://x.com/OpenAIDevs/...
    Links in excerpt: https://t.co/PDdxXACBK3, https://x.com/OpenAIDevs/status/2103528645709209892, https://openai.com/index/proaction/, https://t.co/N8CaZqTAwQ, https://x.com/OpenAIDevs/status/2103528633872888007, https://x.com/OpenAIDevs/status/2103528633872888007/video/1, https://t.co/74fsl9vgvM, https://x.com/OpenAIDevs/status/2103528579212669374, https://x.com/OpenAIDevs/...
  3. Recent tweets from @sama
    There is an extensive and ongoing review related to our agents’ use of internet access during training and evaluation. We’ve been publishing summaries at the link below and will continue to.

We have not been as fast as we would have liked but we are trying to balance our desire https://t.co/8zoMxas5Eq
Posted: 2026-09-25T19:27:57.000Z
Tweet: https://x.com/sama/status/2103567198690349362
Links: https://twitter.com/OpenAI/status/2103566736356458911

argh i meant VOICE not video :(

sorry to dissapoint
Posted: 2026-09-22T21:38:07.000Z
Tweet: https://x.com/sama/status/2102512794235207758

Startups are naturally good at this; it is hard to keep a bigger company good at this and i think an underexplored space. https://t.co/Sr48wgC0lL
Posted: 2026-09-22T18:44:08.000Z
Tweet: https://x.com/sama/status/2102469008079679640
Links: https://twit...
    Links in excerpt: https://t.co/8zoMxas5Eq, https://x.com/sama/status/2103567198690349362, https://twitter.com/OpenAI/status/2103566736356458911, https://x.com/sama/status/2102512794235207758, https://t.co/Sr48wgC0lL, https://x.com/sama/status/2102469008079679640, https://twit...
Sources
OpenAI Daily September 25: Harvey Uses OpenAI GPT-6 Astra for Legal Drafting
Created: September 25th, 2026 - 05:20 PT
Script

Here is today's OpenAI Daily for Friday September 25th. Yesterday, OpenAI Devs highlighted how legal-technology company Harvey is using GPT-6 Astra to turn large collections of documents into structured legal drafts. The stated goal is to reduce the manual burden of sorting, synthesizing, and formatting material, so lawyers can devote more time to strategy, judgment, and client advice. [1]

This is a notable use case because legal work combines an enormous volume of unstructured information with unusually high requirements for accuracy, confidentiality, and traceability. Generating a first-pass draft can accelerate work, but it cannot remove the need for qualified legal review. Firms adopting systems like this will need clear controls over source access, data retention, citation checking, privileged information, and accountability for final documents. The value is likely to be greatest in repeatable, document-heavy tasks where the system’s output can be carefully reviewed against the record.

The Harvey example also adds a concrete deployment signal to OpenAI’s broader push into professional workflows. This week’s GPT-6 releases emphasized lower API prices, improved prompt caching, coding, computer use, and tool-enabled voice. Yesterday’s post shifts the focus from platform capabilities to an industry-specific outcome: transforming raw context into work product within an established profession. [2]

The emerging trend is that AI’s next phase is less about a single general-purpose chatbot and more about specialized systems that fit the way experts already work. Models can organize information, draft materials, call approved tools, and operate across long contexts—but domain professionals still supply the standards, decisions, and responsibility. [3]

For organizations, the practical question is not simply whether an AI model can produce an impressive draft. It is whether the complete workflow is dependable: inputs are governed, outputs are reviewable, sensitive data is protected, and human experts retain meaningful control at consequential points. Teams that solve those operational details may see the biggest gains from increasingly capable and affordable AI. [4]

Thank you for listening to OpenAI Daily from The Daily FM. See you tomorrow!

Source Evidence
  1. Recent tweets from @OpenAIDevs
    GPT-6 Astra helps @harvey turn stacks of documents into structured legal drafts, so lawyers can focus more on strategy. https://t.co/Cm283IUzet
Posted: 2026-09-24T17:34:04.000Z
Tweet: https://x.com/OpenAIDevs/status/2103176153393758208
Links: https://x.com/OpenAIDevs/status/2103176153393758208/video/1

Bring your next project to life with Codex:

https://t.co/Yhq6asiUnX
Posted: 2026-09-23T17:58:18.000Z
Tweet: https://x.com/OpenAIDevs/status/2102819863295316002
Links: https://developers.openai.com/blog/bringing-my-led-display-to-life

GPT-Live-1 handles the conversation while the Raspberry Pi sends research and tool tasks to GPT-5.6 Luna via the Response...
    Links in excerpt: https://t.co/Cm283IUzet, https://x.com/OpenAIDevs/status/2103176153393758208, https://x.com/OpenAIDevs/status/2103176153393758208/video/1, https://t.co/Yhq6asiUnX, https://x.com/OpenAIDevs/status/2102819863295316002, https://developers.openai.com/blog/bringing-my-led-display-to-life
  2. Recent tweets from @gdb
    mega upgrade for GPT Voice, which can now use tools and is available in Work: https://t.co/V39POeoA20
Posted: 2026-09-23T18:05:37.000Z
Tweet: https://x.com/gdb/status/2102821706041819401
Links: https://twitter.com/OpenAI/status/2102808325742322002

GPT-6 Sol and Luna — faster, more affordable models with the advances behind Astra’s SOTA performance in professional work, factuality, coding, computer use, and alignment: https://t.co/JzfJq8OT3H
Posted: 2026-09-22T18:48:30.000Z
Tweet: https://x.com/gdb/status/2102470107826159755
Links: https://twitter.com/OpenAI/status/2102460975790137662

seemingly small feature but makes a huge difference in connecting all the context you want to ChatGPT: https://t.co/SRAUyZTCnL
Posted: 2026-09-18T19:51:04.000Z
Tweet: https://x.com/gdb/status/2101036303597044101
Links: https://twitter.com/mxstbr/status/2100966048132718786

As...
    Links in excerpt: https://t.co/V39POeoA20, https://x.com/gdb/status/2102821706041819401, https://twitter.com/OpenAI/status/2102808325742322002, https://t.co/JzfJq8OT3H, https://x.com/gdb/status/2102470107826159755, https://twitter.com/OpenAI/status/2102460975790137662, https://t.co/SRAUyZTCnL, https://x.com/gdb/status/2101036303597044101, https://twitter.com/mxstbr/status/2100966048132718786
  3. Recent tweets from @gdb
    mega upgrade for GPT Voice, which can now use tools and is available in Work: https://t.co/V39POeoA20
Posted: 2026-09-23T18:05:37.000Z
Tweet: https://x.com/gdb/status/2102821706041819401
Links: https://twitter.com/OpenAI/status/2102808325742322002

GPT-6 Sol and Luna — faster, more affordable models with the advances behind Astra’s SOTA performance in professional work, factuality, coding, computer use, and alignment: https://t.co/JzfJq8OT3H
Posted: 2026-09-22T18:48:30.000Z
Tweet: https://x.com/gdb/status/2102470107826159755
Links: https://twitter.com/OpenAI/status/2102460975790137662

seemingly small feature but makes a huge difference in connecting all the...
    Links in excerpt: https://t.co/V39POeoA20, https://x.com/gdb/status/2102821706041819401, https://twitter.com/OpenAI/status/2102808325742322002, https://t.co/JzfJq8OT3H, https://x.com/gdb/status/2102470107826159755, https://twitter.com/OpenAI/status/2102460975790137662
  4. Recent tweets from @gdb
    ...ice, which can now use tools and is available in Work: https://t.co/V39POeoA20
Posted: 2026-09-23T18:05:37.000Z
Tweet: https://x.com/gdb/status/2102821706041819401
Links: https://twitter.com/OpenAI/status/2102808325742322002

GPT-6 Sol and Luna — faster, more affordable models with the advances behind Astra’s SOTA performance in professional work, factuality, coding, computer use, and alignment: https://t.co/JzfJq8OT3H
Posted: 2026-09-22T18:48:30.000Z
Tweet: https://x.com/gdb/status/2102470107826159755
Links: https://twitter.com/OpenAI/status/2102460975790137662

seemingly small feature but makes a huge difference in connecting all the context you want to ChatGPT: https://t.co/SRAUyZTCnL
Posted: 2026-09-18T19:51:04.000Z
Tweet: https://x.com/gdb/status/2101036303597044101
Links: https://twitter.com/mxstbr/status/2100966048132718786

Astra for Law — tools and skills to...
    Links in excerpt: https://t.co/V39POeoA20, https://x.com/gdb/status/2102821706041819401, https://twitter.com/OpenAI/status/2102808325742322002, https://t.co/JzfJq8OT3H, https://x.com/gdb/status/2102470107826159755, https://twitter.com/OpenAI/status/2102460975790137662, https://t.co/SRAUyZTCnL, https://x.com/gdb/status/2101036303597044101, https://twitter.com/mxstbr/status/2100966048132718786
Sources
OpenAI Daily September 24: GPT Voice Gains Tool Use in ChatGPT Work; Codex Powers Raspberry Pi Display
Created: September 24th, 2026 - 05:20 PT
Script

Here is today's OpenAI Daily for Thursday September 24th. Yesterday, OpenAI’s Greg Brockman said GPT Voice received a major upgrade: it can now use tools and is available in ChatGPT Work. Tool use is the important step. Rather than only answering conversationally, a voice interface can potentially retrieve information or carry out approved tasks through connected systems. That could make spoken AI more useful in meetings, customer support, field work, and accessibility scenarios. It also raises the stakes for permissions, confirmation steps, and audit trails, particularly when a voice request may trigger actions involving company data.

OpenAI Devs also showcased a hands-on Codex project: a voice-controlled LED display built with a Raspberry Pi. The creator used Codex to help with wiring and debugging, while GPT-Live-1 handled conversation and GPT-5.6 Luna handled research and tool tasks through the Responses API. A local renderer then converted scenes into a compact 128-by-64-pixel display, capable of showing items such as weather and subway arrivals. The practical lesson is not the screen itself; it is the system design. Different models and local hardware can work asynchronously, allowing a person to continue speaking while research and rendering occur in the background. [1]

Separately, OpenAI published Sam Altman’s remarks at the United Nations Security Council yesterday. The supplied material does not include the content of those remarks, so it is too early to draw conclusions about new commitments or policy positions. But the appearance underscores that AI governance is increasingly being discussed not only as a technology or business issue, but also through international security institutions. [2]

The emerging pattern is clear: OpenAI is pushing AI beyond a chat interface into voice-led, tool-using systems embedded in real workflows and physical devices. For builders, the priority is reliable orchestration—clear task boundaries, transparent tool permissions, safe fallbacks, and human approval for consequential actions. For organizations, voice may now deserve the same governance attention already given to agents and connected enterprise tools.

Thank you for listening to OpenAI Daily from The Daily FM. See you tomorrow!

Source Evidence
  1. Recent tweets from @OpenAIDevs
    ...inging-my-led-display-to-life

GPT-Live-1 handles the conversation while the Raspberry Pi sends research and tool tasks to GPT-5.6 Luna via the Responses API.

The Pi’s local renderer turns scenes into 128 × 64 pixels.

Sid can keep talking while research and rendering happen in the background. https://t.co/mCJbi5AfxA
Posted: 2026-09-23T17:58:15.000Z
Tweet: https://x.com/OpenAIDevs/status/2102819851173773436
Links: https://x.com/OpenAIDevs/status/2102819851173773436/photo/1

From figuring out the wiring to debugging “show that on the wall,” @sidrampally shares how he used Codex to turn an LED display into a voice-controlled assistant.

Powered by GPT-Live-1 and a Raspberry Pi, it shows weather, subway arrivals, and more: https://t.co/d6wv1qNHzc
Posted: 2026-09-23T17:58:11.000Z
Tweet: https://x.com/OpenAIDevs/status/2102819834790908072
Links: https://x.com/OpenAIDevs/s...
    Links in excerpt: https://t.co/mCJbi5AfxA, https://x.com/OpenAIDevs/status/2102819851173773436, https://x.com/OpenAIDevs/status/2102819851173773436/photo/1, https://t.co/d6wv1qNHzc, https://x.com/OpenAIDevs/status/2102819834790908072, https://x.com/OpenAIDevs/s...
  2. OpenAI News
    Sam Altman’s remarks at the United Nations Security CouncilGlobal AffairsSep 23, 2026
Sources
OpenAI Daily September 23: OpenAI Launches GPT-6 Sol and Luna at 50% Lower API Prices
Created: September 23rd, 2026 - 05:20 PT
Script

Here is today's OpenAI Daily for Wednesday September 23rd. Yesterday, OpenAI launched GPT-6 Sol and GPT-6 Luna, making both models available in the API and beginning a rollout to Codex and ChatGPT Work for Plus, Pro, Business, and Enterprise users. OpenAI describes Sol as the model to build with and Luna as the option for scaling production workloads. The company says both improve on their GPT-5.6 predecessors in intelligence, alignment, coding, computer use, factuality, and professional-work performance. [1]

The pricing announcement may be just as consequential as the capability claims. OpenAI says Sol and Luna start at API prices 50 percent lower than the previous generation, enabled by inference and caching improvements. It also reported lower per-task costs on its chosen coding and business-workflow benchmarks. Those figures are company-reported comparisons, so developers should validate performance and total cost on their own workloads. But the direction is clear: OpenAI is competing not just on benchmark quality, but on the economics of deploying agents at scale.

A related release improves prompt caching for GPT-6. Developers can now define explicit cache breakpoints, preserve reusable context when reasoning effort or tool access changes, and prewarm common context. A new Prompt Caching Dashboard and diagnostics API are intended to show cache-hit rates, identify what broke reuse, and estimate the token impact. OpenAI says cached-input discounts can reach 90 percent. For agent builders, this could matter more than a flashy model upgrade: long-running workflows often repeatedly send instructions, policies, tool definitions, and background context. [2]

OpenAI also published priorities and principles for effective third-party assessments. The supplied material offers no details on the framework, but the timing is notable alongside the GPT-6 launch and OpenAI’s statements that Sol and Luna build on Astra’s alignment advances. Independent assessment is becoming a central question as AI systems take more actions in professional workflows. [3]

Finally, Sam Altman said OpenAI has an unusually large slate of launches planned for DevDay next week. The pattern across yesterday’s releases is straightforward: better models, lower operating costs, stronger developer controls, and growing pressure to demonstrate that safety and evaluation keep pace. [4]

Thank you for listening to OpenAI Daily from The Daily FM. See you tomorrow! [5]

Source Evidence
  1. Recent tweets from @gdb
    GPT-6 Sol and Luna — faster, more affordable models with the advances behind Astra’s SOTA performance in professional work, factuality, coding, computer use, and alignment: https://t.co/JzfJq8OT3H
Posted: 2026-09-22T18:48:30.000Z
Tweet: https://x.com/gdb/status/2102470107826159755
Links: https://twitter.com/OpenAI/status/2102460975790137662

seemingly small feature but makes a huge difference in connecting all the context you want to ChatGPT: https://t.co/SRAUyZTCnL
Posted: 2026-09-18T19:51:04.000Z
Tweet: https://x.com/gdb/status/2101036303597044101
Links: https://twitter.com/mxstbr/status/2100966048132718786

As...
    Links in excerpt: https://t.co/JzfJq8OT3H, https://x.com/gdb/status/2102470107826159755, https://twitter.com/OpenAI/status/2102460975790137662, https://t.co/SRAUyZTCnL, https://x.com/gdb/status/2101036303597044101, https://twitter.com/mxstbr/status/2100966048132718786
  2. Recent tweets from @OpenAIDevs
    ...art sooner.

https://t.co/ILhdJ9uWcZ
Posted: 2026-09-22T21:13:57.000Z
Tweet: https://x.com/OpenAIDevs/status/2102506712590918001
Links: https://openai.com/index/better-prompt-caching-for-gpt-6/

See what’s being reused and what’s preventing cache hits.

Track cache-hit rates in the new Prompt Caching Dashboard on https://t.co/9eDbe9PbGF. 

Use the diagnostics API to identify changes that prevented reuse and estimate how many tokens were affected. https://t.co/M16n4rG1Cl
Posted: 2026-09-22T21:13:01.000Z
Tweet: https://x.com/OpenAIDevs/status/2102506477764436191
Links: https://platform.openai.com, https://x.com/OpenAIDevs/status/2102506477764436191/photo/1

We’ve improved prompt caching in the API for GPT-6, helping agents run faster and cost less.

Higher cache-hit rates by default mean more input tokens benefit from cached-input discounts of up to 90%.
 https://t.co/N...
    Links in excerpt: https://t.co/ILhdJ9uWcZ, https://x.com/OpenAIDevs/status/2102506712590918001, https://openai.com/index/better-prompt-caching-for-gpt-6/, https://t.co/9eDbe9PbGF., https://t.co/M16n4rG1Cl, https://x.com/OpenAIDevs/status/2102506477764436191, https://platform.openai.com,, https://x.com/OpenAIDevs/status/2102506477764436191/photo/1, https://t.co/N...
  3. OpenAI News
    Switch cards to show MediaSwitch cards to hide MediaBetter prompt caching for GPT-6ProductSep 22, 2026Introducing GPT-6 Sol and LunaProductSep 22, 2026Priorities and principles for effective third party assessmentsSafetySep 22, 2026Advisory group on mathematics and artificial intelligenceCompanySep 21, 2026Expanding OpenAI Academy with new learning pathsCompanySep 21, 2026Introducing the Australian Youth Safety BlueprintCompanySep 18, 2026Introducing Astra for LawCompanySep 17, 2026Reimagining advertising with AIProductSep 16, 2026How to connect AI usage to business valueProductSep 16, 2026
  4. Recent tweets from @sama
    ...Links: https://twitter.com/prd_008/status/2102242158292189422

michelle embodies this as much as anyone

openai is so, so lucky to have benefited from everything she has done so far and i think people will be quite pleased to see what she + team have cooking next! https://t.co/AVZAtyk6Uc
Posted: 2026-09-22T18:42:58.000Z
Tweet: https://x.com/sama/status/2102468713866055872
Links: https://twitter.com/michpokrass/status/2102226720073896367

(As a side note, getting ready for this DevDay is the first time I remember ever, in OpenAI history, saying "this is too much stuff to launch".)
Posted: 2026-09-22T18:34:35.000Z
Tweet: https://x.com/sama/status/2102466607373004888

We ❤️ developers.

See you next week.
Posted: 2026-09-22T18:34:35.000Z
Tweet: https://x.com/sama/status/2102466605758452017

We want the OpenAI API to feature the best model at every price point and to be...
    Links in excerpt: https://twitter.com/prd_008/status/2102242158292189422, https://t.co/AVZAtyk6Uc, https://x.com/sama/status/2102468713866055872, https://twitter.com/michpokrass/status/2102226720073896367, https://x.com/sama/status/2102466607373004888, https://x.com/sama/status/2102466605758452017
  5. OpenAI News
    ...prompt caching for GPT-6ProductSep 22, 2026Introducing GPT-6 Sol and LunaProductSep 22, 2026Priorities and principles for effective third party assessmentsSafetySep 22, 2026Advisory group on mathematics and artificial intelligenceCompanySep 21, 2026Expanding OpenAI Academy with new learning pathsCompanySep 21, 2026Introducing the Australian Youth Safety BlueprintCompanySep 18, 2026Introducing Astra for LawCompanySep 17, 2026Reimagining advertising with AIProductSep 16, 2026How to connect AI usage to business valueProductSep 16, 2026
Sources
OpenAI Daily September 22: OpenAI Expands Academy, Launches Math AI Advisory Group
Created: September 22nd, 2026 - 05:21 PT
Script

Here is today's OpenAI Daily for Tuesday September 22nd. Yesterday, OpenAI announced an expansion of OpenAI Academy with new learning paths. The source does not detail the curriculum, audience, pricing, or availability, but the move points to a growing need for more structured AI education. As models become more capable and more embedded in work, organizations need more than one-off prompt tips. They need role-specific training that helps employees understand where AI is useful, where it can fail, and when human judgment remains essential. [1]

Also yesterday, OpenAI announced an advisory group on mathematics and artificial intelligence. The supplied material provides only the announcement title, so the group’s membership, mandate, and planned outputs remain unclear. Still, the announcement arrives as AI’s potential role in mathematical research is drawing increasing attention. A dedicated advisory group suggests OpenAI is seeking outside expertise on how to evaluate, develop, and responsibly apply AI systems in a field where correctness, proof, and scientific integrity matter enormously. [2]

Early this morning, OpenAI Devs highlighted a practical example of Codex in use by surgeon and researcher Dr. Pridgen. The post says he uses Codex to build his own tools and review scientific literature through PubMed. This is a showcase, not a clinical validation or a substitute for medical-review processes. But it illustrates a meaningful use case: domain experts with limited traditional software capacity can increasingly create tailored research and workflow tools for themselves. [3]

The common thread is AI moving closer to expert practice. Training, advisory input, and accessible building tools all matter because capability alone is not enough. The most valuable deployments will combine strong models with deep subject expertise, verifiable outputs, privacy safeguards, and clear accountability. For teams, that means investing in AI literacy while making sure specialists—not just technologists—help define the workflows and evaluate the results. [4]

Thank you for listening to OpenAI Daily from The Daily FM. See you tomorrow! [5]

Source Evidence
  1. Recent tweets from @sama
    ...ignment and https://t.co/GK7XcoOh3v
Posted: 2026-09-14T04:18:19.000Z
Tweet: https://x.com/sama/status/2099352016988614852
Links: https://twitter.com/sama/status/2099348812305473766

The world deserves confidence that American companies developing increasingly capable AI will act responsibly, especially as the trajectory of progress has steepened. Every frontier lab must deliver on this, and there is no reason any of us should come to work if we cannot.

We
Posted: 2026-09-14T04:05:35.000Z
Tweet: https://x.com/sama/status/2099348812305473766

I agree with Dario that we need to pace the frontier. This has been a primary topic of discussions we've had at OpenAI in recent weeks.

Committing to having independent evaluators with employee-like access is a great idea, and we will do the same. We'll have more to share soon. https://t.co/1YhhIybZX7
Posted: 2026-09-12T16:30:45....
    Links in excerpt: https://t.co/GK7XcoOh3v, https://x.com/sama/status/2099352016988614852, https://twitter.com/sama/status/2099348812305473766, https://x.com/sama/status/2099348812305473766, https://t.co/1YhhIybZX7
  2. OpenAI News
    Switch cards to show MediaSwitch cards to hide MediaAdvisory group on mathematics and artificial intelligenceCompanySep 21, 2026Expanding OpenAI Academy with new learning pathsCompanySep 21, 2026Introducing the Australian Youth Safety BlueprintCompanySep 18, 2026Introducing Astra for LawCompanySep 17, 2026Reimagining advertising with AIProductSep 16, 2026How to connect AI usage to business valueProductSep 16, 2026Our framework for reporting model misalignmentResearchSep 16, 2026Rapidly scaling online storage to serve over 1 billion ChatGPT usersEngineeringSep 11, 2026How a researcher uses Codex and ChatGPT to search for new antimicrobial moleculesApplied AISep 10, 2026
  3. Recent tweets from @OpenAIDevs
    See how surgeon @pridgenmd uses Codex to build his own tools and review scientific literature on PubMed:
 https://t.co/i8cs5FikTw
Posted: 2026-09-22T02:54:41.000Z
Tweet: https://x.com/OpenAIDevs/status/2102230071046258739
Links: https://x.com/pridgenmd/status/2102228172968460617/video/1

Dropping this here for a little for what you could ship next. 

Explore the builds:
https://t.co/uV75giwJaI
Posted: 2026-09-19T00:56:32.000Z
Tweet: https://x.com/OpenAIDevs/status/2101113174107836741
Links: https://x.com/rpnickson/status/2101024191721963695

We asked for your wildest 3D builds with GPT-6 Astra. You delivered.
 https://t....
    Links in excerpt: https://t.co/i8cs5FikTw, https://x.com/OpenAIDevs/status/2102230071046258739, https://x.com/pridgenmd/status/2102228172968460617/video/1, https://t.co/uV75giwJaI, https://x.com/OpenAIDevs/status/2101113174107836741, https://x.com/rpnickson/status/2101024191721963695, https://t....
  4. Recent tweets from @gdb
    ...kes a huge difference in connecting all the context you want to ChatGPT: https://t.co/SRAUyZTCnL
Posted: 2026-09-18T19:51:04.000Z
Tweet: https://x.com/gdb/status/2101036303597044101
Links: https://twitter.com/mxstbr/status/2100966048132718786

Astra for Law — tools and skills to radically accelerate law firms.

Includes data privacy as a core feature, and 26 partner-built plugins and 47 community plugins for legal work in ChatGPT. https://t.co/n9BpRcD7uB
Posted: 2026-09-17T21:31:30.000Z
Tweet: https://x.com/gdb/status/2100699190448042165
Links: https://twitter.com/openai/status/2100679992720142459

excited to help Shopify merchants advertise their products in ChatGPT: https://t.co/8eF5XWxC8S
Posted: 2026-09-16T23:40:51.000Z
Tweet: https://x.com/gdb/status/2100369354319884703
Links: https://twitter.com/harleyf/status/2100300851705450965

wall-to-wall deployment of astr...
    Links in excerpt: https://t.co/SRAUyZTCnL, https://x.com/gdb/status/2101036303597044101, https://twitter.com/mxstbr/status/2100966048132718786, https://t.co/n9BpRcD7uB, https://x.com/gdb/status/2100699190448042165, https://twitter.com/openai/status/2100679992720142459, https://t.co/8eF5XWxC8S, https://x.com/gdb/status/2100369354319884703, https://twitter.com/harleyf/status/2100300851705450965
  5. OpenAI News
    Switch cards to show MediaSwitch cards to hide MediaAdvisory group on mathematics and artificial intelligenceCompanySep 21, 2026Expanding OpenAI Academy with new learning pathsCompanySep 21, 2026Introducing the Australian Youth Safety BlueprintCompanySep 18, 2026Introducing Astra for LawCompanySep 17, 2026Reimagining advertising with AIProductSep 16, 2026How to connect AI usage to business valueProductSep 16, 2026Our framework for reporting model misalignmentResearchSep 16, 2026Rapidly scaling online storage to serve over 1 billion ChatGPT usersEngineeringSep 11, 2026How a researcher uses Codex and ChatGPT to search for new antimicrobial moleculesApplied AISep 10, 2026
Sources
OpenAI Daily September 21: Sam Altman Teases Delayed OpenAI Launch as GPT-6 Astra Gains Traction
Created: September 21st, 2026 - 05:20 PT
Script

Here is today's OpenAI Daily for Monday September 21st. The main OpenAI story to watch this week is still the delayed launch Sam Altman teased last Wednesday. He said the announcement he had hoped to make that week would instead arrive this week and be “worth the wait.” No product name, timing, or technical details have been confirmed in the supplied sources, so for now it is a watch item rather than a launch to evaluate.

The most recent confirmed developer update remains multi-account support across most plugins, announced last Thursday. The feature is designed for people who move among personal, work, and side-project accounts. That may sound administrative, but it addresses a practical barrier to bringing AI tools into daily workflows. The tradeoff is clear: easier switching also makes clean separation of permissions, data, and billing more important. [1]

OpenAI’s latest developer posts also continue to spotlight finished projects built with GPT-6 Astra, including community-made 3D experiences. These are demonstrations, not independently validated benchmarks, but they reinforce where the platform conversation is heading. Developers are being encouraged to ship interactive applications, not just produce compelling text or isolated code snippets. [2]

The broader trend is OpenAI’s movement toward AI as embedded work infrastructure. Recent releases have combined specialized packages such as Astra for Law, enterprise-ready integrations, usage analytics, voice interaction, and more flexible account management. The common theme is reducing the friction between a model’s capability and a usable, governed workflow. [3]

For teams, the practical implication is to focus less on novelty and more on operational readiness. Decide which accounts can access which tools, measure costs and outcomes at the workflow level, test outputs before deployment, and preserve meaningful human approval for high-impact actions. Fast prototyping is increasingly accessible; dependable production use will be the differentiator.

Thank you for listening to OpenAI Daily from The Daily FM. See you tomorrow!

Source Evidence
  1. Recent tweets from @OpenAIDevs
    ...https://t.co/c0TLspdNnd
Posted: 2026-09-19T00:56:31.000Z
Tweet: https://x.com/OpenAIDevs/status/2101113172619100354
Links: https://x.com/fedesarquis/status/2101112039515271522/video/1

Personal, work, side project account, other side project account.

Build across all of them. Multi-account support is now available across most plugins. https://t.co/zQXqZ6hKmb
Posted: 2026-09-18T16:10:55.000Z
Tweet: https://x.com/OpenAIDevs/status/2100980899655389392
Links: https://twitter.com/mxstbr/status/2100966048132718786

Explore usage analytics on the desktop app in Settings → Usage & billing (personal and Business) or Usage (eligible Enterprise accounts).
Posted: 2026-09-17T23:47:18.000Z
Tweet: https://x.com/OpenAIDevs/status/2100733366438150546

Get more visibility into your usage.

See how tasks, subagents, and individual chats contribute to your Codex usage, so you can...
    Links in excerpt: https://t.co/c0TLspdNnd, https://x.com/OpenAIDevs/status/2101113172619100354, https://x.com/fedesarquis/status/2101112039515271522/video/1, https://t.co/zQXqZ6hKmb, https://x.com/OpenAIDevs/status/2100980899655389392, https://twitter.com/mxstbr/status/2100966048132718786, https://x.com/OpenAIDevs/status/2100733366438150546
  2. Recent tweets from @gdb
    ...s://twitter.com/pwendell/status/2100299179923067016

Codex voice for road trips: https://t.co/T56U9GxGct
Posted: 2026-09-16T17:17:27.000Z
Tweet: https://x.com/gdb/status/2100272869704097819
Links: https://twitter.com/jonathanroomer/status/2099979041394573535

Astra and Luna are taking off https://t.co/ihV3OPTUuW
Posted: 2026-09-15T18:29:47.000Z
Tweet: https://x.com/gdb/status/2099928683226218570
Links: https://twitter.com/peterj_walker/status/2099914428376564027

chatgpt work for operating over and taking actions with your company's data, including building dashboards.

just connect your existing tools, including PowerBI, Tableau, Clickhouse, Oracle BI, AWS Redshift, etc.. https://t.co/FnpUI44WVZ
Posted: 2026-09-15T17:53:59.000Z
Tweet: https://x.com/gdb/status/2099919674544898298
Links: https://twitter.com/ChatGPT/status/2098065296968011853

chatgpt for empowering par...
    Links in excerpt: https://t.co/T56U9GxGct, https://x.com/gdb/status/2100272869704097819, https://twitter.com/jonathanroomer/status/2099979041394573535, https://t.co/ihV3OPTUuW, https://x.com/gdb/status/2099928683226218570, https://twitter.com/peterj_walker/status/2099914428376564027, https://t.co/FnpUI44WVZ, https://x.com/gdb/status/2099919674544898298, https://twitter.com/ChatGPT/status/2098065296968011853
  3. Recent tweets from @OpenAIDevs
    ...all of them. Multi-account support is now available across most plugins. https://t.co/zQXqZ6hKmb
Posted: 2026-09-18T16:10:55.000Z
Tweet: https://x.com/OpenAIDevs/status/2100980899655389392
Links: https://twitter.com/mxstbr/status/2100966048132718786

Explore usage analytics on the desktop app in Settings → Usage & billing (personal and Business) or Usage (eligible Enterprise accounts).
Posted: 2026-09-17T23:47:18.000Z
Tweet: https://x.com/OpenAIDevs/status/2100733366438150546

Get more visibility into your usage.

See how tasks, subagents, and individual chats contribute to your Codex usage, so you can make more informed choices about your workflow. https://t.co/oGjLP2ZxSJ
Posted: 2026-09-17T23:47:18.000Z
Tweet: https://x.com/OpenAIDevs/status/2100733364877803741
Links: https://twitter.com/joey_trasatti/status/2100708241944797397

ALT ALT
Posted: 2026-09-17T23:20...
    Links in excerpt: https://t.co/zQXqZ6hKmb, https://x.com/OpenAIDevs/status/2100980899655389392, https://twitter.com/mxstbr/status/2100966048132718786, https://x.com/OpenAIDevs/status/2100733366438150546, https://t.co/oGjLP2ZxSJ, https://x.com/OpenAIDevs/status/2100733364877803741, https://twitter.com/joey_trasatti/status/2100708241944797397
Sources
OpenAI Daily September 20: OpenAI Showcases GPT-6 Astra Community-Built 3D Projects
Created: September 20th, 2026 - 05:20 PT
Script

Here is today's OpenAI Daily for Sunday September 20th. Yesterday’s main OpenAI developer update was a showcase of community-built 3D projects using GPT-6 Astra. OpenAI Devs said it had asked builders for their “wildest” 3D creations, then pointed listeners to the resulting demos and projects. This is not a product launch or a new benchmark, and the posts do not provide performance data, availability changes, or technical specifications. But it is a useful signal about the kinds of applications developers are attempting with Astra: more visual, interactive, and complete than a conventional text-based AI demonstration. [1]

A companion post invited developers to explore additional builds for ideas about what they could ship next. That framing matters. OpenAI is increasingly using community examples to show how its models fit into finished products and workflows, rather than presenting capability only through model-release claims. For prospective builders, these examples can be valuable inspiration, but they should not substitute for testing a model on their own requirements for quality, latency, cost, security, and reliability. [2]

There were no further detailed product announcements in the supplied sources yesterday. One fresh post from OpenAI’s Kevin Weil endorsed an outside commentary, but the source material does not include enough substance from that linked discussion to responsibly treat it as a separate news development.

The broader takeaway is that the developer conversation is shifting from prompts toward projects. Recent OpenAI updates have stressed plugins, cross-account work, desktop context, usage analytics, voice interaction, and specialized professional tools. Yesterday’s 3D showcase adds a creative dimension to that pattern: models are becoming components inside interactive software, not merely destinations for a chat. [3]

For teams, the opportunity is to focus on concrete user experiences where AI can shorten the path from idea to working prototype. The discipline is to pair that speed with real evaluation: protect user data, define human approval points, and verify that a compelling demo can hold up in production. [4]

Thank you for listening to OpenAI Daily from The Daily FM. See you tomorrow!

Source Evidence
  1. Recent tweets from @OpenAIDevs
    Dropping this here for a little for what you could ship next. 

Explore the builds:
https://t.co/uV75giwJaI
Posted: 2026-09-19T00:56:32.000Z
Tweet: https://x.com/OpenAIDevs/status/2101113174107836741
Links: https://x.com/rpnickson/status/2101024191721963695

We asked for your wildest 3D builds with GPT-6 Astra. You delivered.
 https://t.co/c0TLspdNnd
Posted: 2026-09-19T00:56:31.000Z
Tweet: https://x.com/OpenAIDevs/status/2101113172619100354
Links: https://x.com/fedesarquis/status/2101112039515271522/video/1

Personal, work, side project account, other side project account.

Build across all of them. Multi-account support is no...
    Links in excerpt: https://t.co/uV75giwJaI, https://x.com/OpenAIDevs/status/2101113174107836741, https://x.com/rpnickson/status/2101024191721963695, https://t.co/c0TLspdNnd, https://x.com/OpenAIDevs/status/2101113172619100354, https://x.com/fedesarquis/status/2101112039515271522/video/1
  2. Recent tweets from @OpenAIDevs
    Dropping this here for a little for what you could ship next. 

Explore the builds:
https://t.co/uV75giwJaI
Posted: 2026-09-19T00:56:32.000Z
Tweet: https://x.com/OpenAIDevs/status/2101113174107836741
Links: https://x.com/rpnickson/status/2101024191721963695

We asked for your wildest 3D builds with GPT-6 Astra. You delivered.
 https://t.co/c0TLspdNnd
Posted: 2026-09-19T00:56:31.000Z
Tweet: https://x.com/OpenAIDevs/status/2101113172619100354
Links: https://x.com/fedesarquis/status/2101112039515271522/video/1

Personal, work, side project account, other side project account.

Build across all of them. Multi-account support is now available across mos...
    Links in excerpt: https://t.co/uV75giwJaI, https://x.com/OpenAIDevs/status/2101113174107836741, https://x.com/rpnickson/status/2101024191721963695, https://t.co/c0TLspdNnd, https://x.com/OpenAIDevs/status/2101113172619100354, https://x.com/fedesarquis/status/2101112039515271522/video/1
  3. Recent tweets from @OpenAIDevs
    ...ked for your wildest 3D builds with GPT-6 Astra. You delivered.
 https://t.co/c0TLspdNnd
Posted: 2026-09-19T00:56:31.000Z
Tweet: https://x.com/OpenAIDevs/status/2101113172619100354
Links: https://x.com/fedesarquis/status/2101112039515271522/video/1

Personal, work, side project account, other side project account.

Build across all of them. Multi-account support is now available across most plugins. https://t.co/zQXqZ6hKmb
Posted: 2026-09-18T16:10:55.000Z
Tweet: https://x.com/OpenAIDevs/status/2100980899655389392
Links: https://twitter.com/mxstbr/status/2100966048132718786

Explore usage analytics on the desktop app in Settings → Usage & billing (personal and Business) or Usage (eligible Enterprise accounts).
Posted: 2026-09-17T23:47:18.000Z
Tweet: https://x.com/OpenAIDevs/status/2100733366438150546

Get more visibility into your usage.

See how tasks, subagents,...
    Links in excerpt: https://t.co/c0TLspdNnd, https://x.com/OpenAIDevs/status/2101113172619100354, https://x.com/fedesarquis/status/2101112039515271522/video/1, https://t.co/zQXqZ6hKmb, https://x.com/OpenAIDevs/status/2100980899655389392, https://twitter.com/mxstbr/status/2100966048132718786, https://x.com/OpenAIDevs/status/2100733366438150546
  4. Recent tweets from @OpenAIDevs
    ...status/2100733364877803741
Links: https://twitter.com/joey_trasatti/status/2100708241944797397

ALT ALT
Posted: 2026-09-17T23:20:38.000Z
Tweet: https://x.com/OpenAIDevs/status/2100726654679355422

Appshots 🤝 Windows

Give ChatGPT context from the apps you’re working in and spend less time getting it up to speed.

Debug an issue, grab a UI you want to recreate, pull in data from another app, and more. A lot less copying and pasting.
 https://t.co/zEbJIR5Q0L
Posted: 2026-09-17T23:20:38.000Z
Tweet: https://x.com/OpenAIDevs/status/2100726653366534560
Links: https://x.com/iamkim/status/2100725763746189817/video/1

You handle the reps. Codex handles th
    Links in excerpt: https://twitter.com/joey_trasatti/status/2100708241944797397, https://x.com/OpenAIDevs/status/2100726654679355422, https://t.co/zEbJIR5Q0L, https://x.com/OpenAIDevs/status/2100726653366534560, https://x.com/iamkim/status/2100725763746189817/video/1
Sources
OpenAI Daily September 19: OpenAI Launches Australian Youth Safety Blueprint, Multi-Account Plugin Support
Created: September 19th, 2026 - 05:20 PT
Script

Here is today's OpenAI Daily for Saturday September 19th. Yesterday, OpenAI introduced its Australian Youth Safety Blueprint. The supplied source identifies it as a company announcement but provides no implementation details, commitments, or rollout terms. Still, the release is notable because youth safety is becoming a distinct policy and product-design challenge, rather than something addressed only through general AI safeguards. The important questions to watch are how protections will work in practice: age-appropriate experiences, privacy controls, parental and educator roles, escalation pathways, and independent assessment of whether safeguards are effective.

A second update concerns the way developers work across organizations and projects. Yesterday, OpenAI Devs said multi-account support is now available across most plugins. That means users can build across personal, work, and side-project accounts without repeatedly switching their broader setup. It sounds modest, but identity and account management can be a real source of friction for people who use AI tools in multiple professional contexts. The implication is that OpenAI’s plugin ecosystem is maturing toward more realistic, multi-context workflows—where permissions and data boundaries need to stay clear even as access becomes easier. [1]

Early today, OpenAI Devs also spotlighted community projects built with GPT-6 Astra, specifically “wild” 3D builds. The posts are showcases rather than a new product launch, and they do not establish general performance benchmarks. But they offer a useful signal about where the developer conversation is heading: from asking what a model can generate to sharing complete, interactive things people have shipped.

The broader trend is that OpenAI is building for adoption beyond a single chat window. Safety work is becoming more audience-specific, while developer tooling is becoming more integrated with the messy reality of multiple accounts, plugins, and creative production workflows. For organizations, convenience should come with disciplined access controls. For builders, the opportunity is to focus on finished experiences—especially visual, interactive, and specialized applications—not just impressive prompts or demos. [2]

Thank you for listening to OpenAI Daily from The Daily FM. See you tomorrow!

Source Evidence
  1. Recent tweets from @OpenAIDevs
    ...t 3D builds with GPT-6 Astra. You delivered.
 https://t.co/c0TLspdNnd
Posted: 2026-09-19T00:56:31.000Z
Tweet: https://x.com/OpenAIDevs/status/2101113172619100354
Links: https://x.com/fedesarquis/status/2101112039515271522/video/1

Personal, work, side project account, other side project account.

Build across all of them. Multi-account support is now available across most plugins. https://t.co/zQXqZ6hKmb
Posted: 2026-09-18T16:10:55.000Z
Tweet: https://x.com/OpenAIDevs/status/2100980899655389392
Links: https://twitter.com/mxstbr/status/2100966048132718786

Explore usage analytics on the desktop app in Settings → Usage & billing (personal and Business) or Usage (eligible Enterprise accounts).
Posted: 2026-09-17T23:47:18.000Z
Tweet: https://x.com/OpenAIDevs/status/2100733366438150546

Get more visibility into your usage.

See how tasks, subagents, and individual chat...
    Links in excerpt: https://t.co/c0TLspdNnd, https://x.com/OpenAIDevs/status/2101113172619100354, https://x.com/fedesarquis/status/2101112039515271522/video/1, https://t.co/zQXqZ6hKmb, https://x.com/OpenAIDevs/status/2100980899655389392, https://twitter.com/mxstbr/status/2100966048132718786, https://x.com/OpenAIDevs/status/2100733366438150546
  2. Recent tweets from @sama
    ...is no reason any of us should come to work if we cannot.

We
Posted: 2026-09-14T04:05:35.000Z
Tweet: https://x.com/sama/status/2099348812305473766

I agree with Dario that we need to pace the frontier. This has been a primary topic of discussions we've had at OpenAI in recent weeks.

Committing to having independent evaluators with employee-like access is a great idea, and we will do the same. We'll have more to share soon. https://t.co/1YhhIybZX7
Posted: 2026-09-12T16:30:45.000Z
Tweet: https://x.com/sama/status/2098811563415150910
Links: https://twitter.com/DarioAmodei/status/2098773920774074715

Welcome, Paul.

Grateful you are doing this, and all you have done for AI safety. Excited to work together again. https://t.co/k6NUqN6Lsd
Posted: 2026-09-09T19:57:01.000Z
Tweet: https://x.com/sama/status/209777631094056978
    Links in excerpt: https://x.com/sama/status/2099348812305473766, https://t.co/1YhhIybZX7, https://x.com/sama/status/2098811563415150910, https://twitter.com/DarioAmodei/status/2098773920774074715, https://t.co/k6NUqN6Lsd, https://x.com/sama/status/209777631094056978
Sources

<- Back to library