TBPN
TBPN

Sep 3, 2026 · 30 min

Listen from 19:55

Listen at 19:55

GPT-6 Astra reignites the fight over AI benchmarks

Model Mayhem, GPT-6 Astra, Why Nvidia Bought Hugging Face | Diet TBPN

The episode captures an industry where model launches, enterprise safeguards, open-source platforms, and shifting evaluations are reshaping who gets trusted and why.

3 key takeaways
  1. 1GPT-6 Astra’s launch turns a striking benchmark result into a new debate over what current evaluations actually measure.
  2. 2Nvidia’s reported Hugging Face acquisition would connect open-source models and developer tools more closely to its hardware ecosystem.
  3. 3Enterprise AI is emerging as a larger commercial opportunity than consumer companions, while unconventional marketing exposes the sector’s excesses.

Don't miss

The hosts react to GPT-6 Astra’s launch and striking ARC AGI 3 result, then examine why the next benchmark must measure what Astra still cannot do.

The brief

John Coogan and Jordi Hays open on a week of rapid model releases, outages, and benchmark disputes, arguing that hands-on testing and trusted reviewers increasingly matter more than standardized scores.

Anthropic’s enterprise safeguards bring a practical tension into focus: companies want misuse monitoring and data sovereignty without sending sensitive customer information into a model provider’s databases.

Nvidia’s reported $12.9303 billion Hugging Face acquisition would give the chipmaker a potential front door to open-source models, frameworks, inference tools, and developers.

The hosts argue that enterprise software offers a much larger market than AI companions, while Fish.audio’s silent billboard becomes a sharp example of technology marketing losing the plot.

OpenAI’s GPT-6 Astra launch delivers a striking ARC AGI 3 result, but the episode’s deeper point is that every apparent benchmark breakthrough creates new tests for what remains unsolved.

What was said on this episode

19 statements · 12 positive · 1 negative · 2 mixed · 4 neutral

  1. John Cooganon Anthropic Fable 5.1Positive2:56

    Anthropic’s Fable 5.1 is strongest among the three newly released models.

    “Anthropic seems to have the strongest model of the three with Fable 5.1 scoring 66 on the artificial intelligence index.”

    Listen at 2:56

  2. John Cooganon Google Gemini 3.8 FlashPositive5:49

    Gemini 3.8 Flash scored 73.7% on DeepSwee and competes with costlier models.

    “It scored 73.7% on DeepSwee, just behind Opus 5, and competitive with models that cost several times more.”

    Listen at 5:49

  3. John Cooganon OpenAI and Anthropic enterprise revenue concentrationNeutral7:02

    OpenAI and Anthropic derive 80% of enterprise revenue from 1% of companies.

    “OpenAI Ananthropic. 80% of their enterprise revenue comes from just 1% of the companies.”

    Listen at 7:02

  4. John Cooganon Top 1% of American companies by salesNeutral8:24

    The top 1% of American companies generate 80% of total revenue.

    “the top 1% of American companies by sales generate 80% of total revenue.”

    Listen at 8:24

  5. John Cooganon Enterprise AI spendingMixed9:15

    Enterprise AI spending may be consumption-based and proportional to company revenue.

    “because enterprise AI particularly, you're not going to be on the $20 plan. You're not going to be on the $200 plan. You're going to be consumption-based, and you're going to look at it a lot more like a marketing line item that's proportional to your revenue, potentially.”

    Listen at 9:15

  6. John Cooganon Nvidia acquisition of Hugging FacePositive10:56

    Nvidia agreed to acquire Hugging Face for approximately $13 billion.

    “NVIDIA is going even bigger on open source with the $13 billion acquisition for hugging face.”

    Listen at 10:56

  7. John Cooganon Nvidia position in AI ecosystemPositive12:40

    Jensen Huang wants Nvidia to maintain dominance in the AI ecosystem.

    “he wants to maintain NVIDIA's dominant position in the AI ecosystem”

    Listen at 12:40

  8. John Cooganon Hugging FacePositive13:07

    Hugging Face functions as the GitHub of AI.

    “Hugging Face is the GitHub of AI.”

    Listen at 13:07

  9. John Cooganon Hugging FacePositive13:32

    Hugging Face has 18 million developers and 200,000 companies using its product.

    “They have over 18 million developers, 200,000 companies using the product, 3 million models, and over half a million data sets.”

    Listen at 13:32

  10. John Cooganon Hugging FacePositive18:59

    Hugging Face became profitable in 2025 while retaining half its raised capital.

    “They became profitable in 2025, still had half the money that they raised.”

    Listen at 18:59

  11. John Cooganon Nvidia offer for Hugging FaceNeutral19:04

    Hugging Face rejected Nvidia’s $500 million offer at a $7 billion valuation.

    “NVIDIA came in to offer 500 million late 2025 at a 7 billion valuation, but they turned it down.”

    Listen at 19:04

  12. John Cooganon Hugging FacePositive19:55

    Hugging Face could funnel open-source users toward Nvidia chips, inference, and compute.

    “if you can be sort of the front door to that and then say, hey, you found your best model, your framework on Hugging Face, and now you're ready to go and buy chips or buy inference or buy compute, and NVIDIA is right there. That is a very logical flow.”

    Listen at 19:55

  13. Jordi Hayson AI romantic companion marketNegative24:08

    The romantic AI companion market does not generate trillion-dollar revenue.

    “there's just not a trillion dollars in revenue.”

    Listen at 24:08

  14. John Cooganon Enterprise software versus AI romantic companionsMixed24:58

    Enterprise software offers more value than AI romantic companions.

    “it makes sense. There's a lot of value in the enterprise, much less in this controversial topic.”

    Listen at 24:58

  15. Jordi Hayson Fish.audioPositive25:53

    Fish.audio provides speech, audio separation, voice-changing, and translation tools.

    “They make text-to-speech, speech-to-text, audio separation, voice changers, translation.”

    Listen at 25:53

  16. OpenAI published the GPT-6 Astra announcement.

    “gpt6 astra has landed the blog post is up”

    Listen at 26:29

  17. John Cooganon OpenAI AstraPositive26:48

    GPT-6 Astra scored 99.9% on ARC AGI 3.

    “the score is 99.9%.”

    Listen at 26:48

  18. John Cooganon OpenAI AstraPositive27:51

    GPT-6 Astra scored 64.6% at maximum reasoning effort and cost $26.

    “GPT-6 Astra scored 64.6% on reasoning effort max, cost 26.”

    Listen at 27:51

  19. John Cooganon OpenAI AstraPositive28:10

    GPT-6 Astra is the world’s best computer-use model.

    “The world's best computer use model”

    Listen at 28:10

Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.

Weekly email

State of the Pod, every Monday

The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.

Listen to the full episode and explore every guest, topic, and moment on PodLume.

GPT-6 Astra reignites the fight over AI benchmarks · PodLume