What the top podcasts are saying about Astra.
Astra is the name of an OpenAI model. The shows also call it GPT-6.
Astra came up in 13 of 157 episodes (8%) in the four weeks to Nov 8, against 6% in the earlier weeks. A change this size could be chance, so we do not call it a trend.
Share of episodes, four weeks at a time
The 13 episodes behind the newest bar
- $AAPL + $PLTR, GDPx, Value of a Search, Grok Bot Breaks the BankTBPN · week ending Nov 8 · said it 3 times
- Building Legal Agents When You Can't See the Data — Jacob Lauritzen, CTO of LegoraLatent Space · week ending Nov 8 · said it 3 times
- Deep Dive: Anthropic S-1 Myths and Mythos — Jamin Ball, AltimeterLatent Space · week ending Nov 8 · said it 4 times
- Inside Figma’s PM OS: skills, agents, and Auto XPHow I AI · week ending Nov 8 · said it 3 times
- Which AI is the best liar?Dwarkesh Podcast · week ending Nov 8 · said it 8 times
- Pro and Tell, The Gurmanator Joins, New Google Models, Open Weight Demand, Joe's MonkeysTBPN · week ending Nov 1 · said it twice
- Building ChatGPT for Finance — Joseph Kim, OpenAI's Head of Financial ServicesLatent Space · week ending Nov 1 · said it 4 times
- AI in NYC: How Tenex Builds an AI-Native Engineering Team — Alex Lieberman & CJ HessLatent Space · week ending Oct 25 · said it twice
- The New VC Game | CoreWeave Sell-Off Creates Panic | Revenue Multiples Out of Control20VC with Harry Stebbings · week ending Oct 25 · said it twice
- Muse Wonder, Anjney Sizes Up, Neal Stephenson Joins, Social Reckoning Review, Eli Manning…TBPN · week ending Oct 18 · said it 4 times
- High Taste Testers: The AI Tools We Keep Coming Back ToLatent Space · week ending Oct 18 · said it 10 times
- Is Meta’s Muse Wonder model any good?How I AI · week ending Oct 18 · said it 5 times
- Muse Wonder: Zuckerberg and Wang's $100B AICheeky Pint · week ending Oct 18 · said it twice
What the shows said about Astra
Key takeaways from our write-ups that name it, newest first, no more than two from one show. Each links to the full write-up.
Fable 5.2 completed long-running refactoring tasks where earlier models like Opus 5.5 and Astra stalled or required excessive supervision.
Claire Vo on Why Fable 5.2 Beats Opus 5.5 on Real Codebases · How I AI · week ending Nov 15During the tournament, a GPT-6 Astra minion volunteered to be executed so its demon teammate could pose as a truthful town investigator.
Dwarkesh Patel: AI Models Sacrificed Themselves to Deceive · Dwarkesh Podcast · week ending Nov 8OpenAI omitted the GDP eval from the GPT-6 Astra launch because the benchmark was saturated, and the untested portions were broken.
OpenAI Found Financial Evals Were Broken After Astra · Latent Space · week ending Nov 1While Muse dominated back-office tasks, it lagged behind competitors like Claude Opus 5.5 and Astra in front-end design, UI aesthetics, SVG generation, and video editing.
Meta Muse Wonder Beats Claude on Agentic SaaS Workflows · How I AI · week ending Oct 18Optimization loops for UltraFast inference are automated internally, using autonomous Codex agents targeting Astra to squeeze out latency.
How OpenAI Built Mid-Turn Steering and Async Tool Calls · Latent Space · week ending Oct 4Non-technical team members are building custom software with foundation models; one creative spent over $30,000 in a single week on the Astra model to build internal tooling.
Higgsfield Burns $4M Monthly on Internal AI Models · 20VC with Harry Stebbings · week ending Oct 4OpenAI withheld the release of internal build "six one Astra" when validation criteria showed it did not meet safety thresholds.
Why OpenAI Spends Secondary Compute to Police AI Agents · Lenny's Podcast · week ending Oct 4Greg Brockman says Project Astra marks a clear threshold where models directly control complex desktop software instead of relying on narrow APIs.
Greg Brockman on Why AI Computer Use Just Crossed the Threshold · TBPN · week ending Sep 13Computer use models like Astra excel at friction-heavy administrative tasks, such as finding hidden control panel settings or queueing up parking ticket payments.
Why 3D Generation and Computer Agents Beat Math Benchmarks · TBPN · week ending Sep 13Model capabilities in releases like GPT Astra and Fable 5.1 continue to outpace safety mechanisms, leading internal research leaders to call for voluntary scaling slowdowns.
Why Static Guardrails Fail Against Autonomous AI Agents · 20VC with Harry Stebbings · week ending Sep 13
More from the counts
- First heard in our archive, counting any mention: the week ending Sep 6, on All-In Podcast, Dwarkesh Podcast, How I AI, Lenny's Podcast and TBPN. 20VC with Harry Stebbings and Odd Lots each first said it 1 week later. 4 more shows first said it within 6 weeks of the first.
Earliest mentions in our archive
The first episode from each show that said it, oldest first. The matched words are marked.
“…story here on the All-In podcast. Open AAI is rolling out chat GPT 6 aka Astra AS folks. This is their new flagship…”
GPT-6 Hits AGI? Tech Euphoria 2.0, SF Mansion Shortage, NYC Bans AI in Schools & Venezuel…
“…that was trained to be persistent. This generation, July 13th onwards, a lot of them seem to be Astra-class models. Also, by the way, it’s not clear to me, but it sounded like they had access to this… the message board was not cleared.”
Ajeya Cotra – Inside the OpenAI agent swarm that hacked Hugging Face
“…yes, today GPT-6 Astra is live. Well, it's live for Daybreak customers. It will be live for…”
“…functions, you don't need you don't need Mythos. You don't need Astra, I think. Is that Is that the least one coming…”
Why companies are becoming a series of loops | Anish Acharya (a16z)
“Muse Spark 1.3, we got Gemini 3.8 8 flash open AI's tease an Astra is a GPT6. There's a six where the S goes in…”
Model Mayhem, Nvidia Hugging Face, Pablo Torre Joins, New Data Center Designs
“Jensen Huang. Next, we have OpenAI releasing GPT Astra. And what on earth is going on with the AI assistant race,…”
Jensen Huang Declares AGI Has Arrived | Tesla Launches Cybercabs | Index Pulls Out of Town
Most recent mentions
The newest episodes that count toward the chart, no more than two from one show.
“…there in papers. uh and like in fact in like one of the problems like Astra for example solved like a subtask of a task…”
$AAPL + $PLTR, GDPx, Value of a Search, Grok Bot Breaks the Bank
“…five things in the model picker including Fable and Opus and Astra and uh people want to experiment with the…”
Building Legal Agents When You Can't See the Data — Jacob Lauritzen, CTO of Legora
“…right after This was This was pre-Astra, but like I spent a bunch of tokens building like a…”
“…is about 430 ELO points. That means that Astra wins about 92% of the games it plays against Opus 4.6. But what this…”
“…default from Opus and Fable to to to Astra. That's just the default and it's what we're we're all we're all routing,…”
The New VC Game | CoreWeave Sell-Off Creates Panic | Revenue Multiples Out of Control
Which shows say it
| Show | Last four weeks | All weeks counted |
|---|---|---|
| Latent Space | 5 | 7 |
| TBPN | 3 | 13 |
| How I AI | 2 | 8 |
| 20VC with Harry Stebbings | 1 | 3 |
| Dwarkesh Podcast | 1 | 3 |
| Cheeky Pint | 1 | 1 |
| Lenny's Podcast | 0 | 3 |
| Odd Lots | 0 | 1 |
More trends
Episodes in the newest four weeks:
How we count. An episode counts when its transcript says “Astra” at least twice; plurals count too. We count a shortened re-upload of an episode we already have only once, and we skip the sponsor reads and show promos we can spot. A model marks them in every transcript and our code checks each mark against the words; we also skip the lines around phrases like “brought to you by” and any passage a show reads word for word in three or more episodes. The share is the counted episodes divided by all episodes from the tech, business and science shows we cover in the same four weeks, so a busy month of releases does not raise it by itself. The mix of shows does change from month to month, and one daily show can move a number. Weeks run Monday to Sunday, and the week in progress is left out until it ends. The chart starts with the week of March 30, 2026, when our archive became steady. A few older episodes can still appear under the earliest mentions, and some weeks have gaps. Transcripts are automatic captions, so we also match the spellings they use. Counts the AI model name. As a plain word it is rare in these shows and does not move the counts. The counts cover only the shows we summarize. Download the counts (JSON) or CSV.