Anthropic's secret Mythos model scored 78 on a coding benchmark where GPT-5.4 scored 58 — and it found a 27-year-old bug that hardened OpenBSD missed for nearly three decades.
God Mode Podcast
Anthropic's secret Mythos model scored 78 on a coding benchmark where GPT-5.4 scored 58 — and it found a 27-year-old bug that hardened OpenBSD missed for nearly three decades.
TL;DR
Rik, Ben, and Luca break down the biggest AI story of the week: Anthropic's leaked Mythos model scores 78 on a coding benchmark where GPT-5.4 scores only 58, and has already found a 27-year-old bug in OpenBSD under Project Glasswing. The crew also covers Anthropic's sudden OAuth kill that evicted OpenClaw users overnight, Google quietly outpacing everyone with free Vids and Gemma 4, OpenAI's $200M TBPN media acquisition, and Intel's rising chip partnerships. Key takeaway: AI intelligence is heading toward commoditization, and the real battle now is ecosystem lock-in.
Rik, Ben, and Luca unpack Anthropic's leaked Mythos model (scoring 78 on a coding benchmark vs GPT-5.4's 58), its Project Glasswing cybersecurity rollout that found a 27-year-old OpenBSD bug, Anthropic's sudden OAuth kill that evicted OpenClaw users, the advisor/executor model strategy, Google's Gemma 4 and free Vids launch, OpenAI's TBPN acquisition, Intel's rising partnerships, and ZAI's GLM 5.1 open-source threat.
Chapter 1 · 00:00
Mythos scored 78. GPT-5.4 scored 58. That gap doesn't happen by accident.
Mythos doesn't just beat GPT-5.4 — it obliterates it. A 20-point coding benchmark gap this large doesn't come from incremental tuning. It suggests Anthropic may have cracked a fundamentally new model architecture beyond the transformer.
On a coding benchmark, Anthropic's Mythos model scored 78 while GPT-5.4 scored 58 — a 20-point gap that suggests a potentially different underlying architecture.
Chapter 3 · 01:15
Mythos found a 27-year bug in OpenBSD. Now imagine an adversary with the same tool.
Mythos, deployed quietly under Project Glasswing to 20 strategic partners including Apple, Google, and Microsoft, found thousands of high-severity vulnerabilities — including a 27-year-old flaw in OpenBSD. If adversaries get a similar model, the banking system and critical infrastructure could be catastrophically exposed.
Anthropic gave Mythos preview access to 20 strategic partners including AWS, Apple, Broadcom, Cisco, Google, and Microsoft before any public release.
Mythos is approximately 50% more capable than Anthropic's previous flagship Opus model, scoring roughly 25 percentage points higher on key benchmarks.
Mythos preview discovered a 27-year-old security vulnerability in OpenBSD, one of the most hardened operating systems ever built.
Chapter 4 · 04:20
Anthropic gave OpenClaw users until 8PM to migrate. That's not a transition — that's an eviction.
Anthropic killed OAuth with a same-day deadline: migrate your OpenClaw agents by 8PM or lose access. Ben had weeks of infrastructure at stake. The move reveals how aggressively Anthropic is trying to lock users into its own managed agents ecosystem before competitors catch up.
Chapter 5 · 06:30
Anthropic gave OpenClaw users until 8PM the same night they received the email to migrate their agents off Claude subscriptions.
Anthropic gave OpenClaw users until 8PM the same night they received the email to migrate their agents off Claude subscriptions.
Anthropic offered displaced OpenClaw users either a full subscription refund or a $200 API credit to migrate their workflows.
Chapter 6 · 09:30
It doesn't hurt to keep your OpenClaw file, like, stored. And you can always sync back to it, like, in six months.
Chapter 7 · 13:20
Why does AI charge businesses MORE than consumers? Luca says it's a sign the market is broken.
In every normal industry, businesses get volume discounts. In AI, the opposite is true — consumers get the cheapest access while businesses pay through the nose on the API. Luca argues this inversion means user loyalty is paper-thin: the moment a competitor matches Anthropic's quality, everyone will migrate.
Anthropic is formalizing what power users already knew: Opus should advise, not execute. Sonnet does the work. Haiku handles the repetitive stuff. This tiered model-routing strategy, now rolled out to all users, could end the epidemic of people maxing out $20 plans in three prompts.
Chapter 8 · 14:40
Anthropic's Opus model costs approximately two to three times more per token than Sonnet, making the new tiered advisor strategy critical for cost management.
Anthropic's Opus model costs approximately two to three times more per token than Sonnet, making the new tiered advisor strategy critical for cost management.
Chapter 9 · 17:40
Jensen Huang says your $250K engineer should be spending $500K on AI. Are they?
Jensen Huang says a $250K engineer who isn't spending $500K on AI API tokens is broken. Meta reportedly has a leaderboard rewarding whoever burns the most tokens. AI spend is no longer overhead — it's the performance metric.
NVIDIA CEO Jensen Huang stated that any $250K/year software engineer not spending at least $500K in AI API token credits is underperforming.
Chapter 10 · 19:35
Anthropic overtook OpenAI in revenue — but Luca says being the leader right now might be a disadvantage.
Anthropic hit a $30B run rate and leapfrogged OpenAI in revenue — despite having a fraction of the users. But Luca flips the narrative: every new user burns compute Anthropic needs to build Mythos. Being the revenue leader while selling credits at a loss may be the slowest path to winning.
Anthropic has reportedly surpassed OpenAI in revenue, hitting a $30 billion annualized run rate.
Rik cited a previous episode's finding that Anthropic trains its models approximately four times more efficiently than OpenAI, meaning it needs only a quarter of the compute.
Chapter 11 · 22:10
Should AI compute cure diseases or make memes? Luca says we're not asking the right question.
Luca asks the question the AI industry keeps avoiding: are we pointing the world's most powerful resource at the right problems? We're automating marketing while diseases go uncured and roads remain unsafe. The conversation has to move from whether to use AI to where it creates the most benefit for humanity.
Chapter 12 · 25:30
OpenAI reportedly projected $100 billion in advertising revenue by 2030, a goal that mirrors Google's ad-based business model.
OpenAI reportedly projected $100 billion in advertising revenue by 2030, a goal that mirrors Google's ad-based business model.
Chapter 13 · 26:40
OpenAI just bought a one-year-old podcast for $200 million. Luca calls it what it is: narrative control.
OpenAI just bought TBPN — a Silicon Valley podcast that had Zuckerberg and Kalanick on within its first year — for $200 million. The crew can't quite figure out why. Luca says it outright: this feels like wanting to control the narrative. Ben counters that at $100B projected ad revenue, $200M is rounding error.
OpenAI acquired the tech podcast and media platform TBPN for approximately $200 million, just one year after TBPN launched.
Chapter 14 · 30:25
Gemma 4 went from 20% to 89% on math in ONE generation. It runs on your phone. Offline.
Gemma 4 didn't just improve — it exploded. Math went from 20% to 89%. Coding from 29% to 80%. It runs offline on a phone. Google is proving that efficient, local AI is the real frontier, and it can be licensed for commercial use — something Anthropic's closed model strategy can't match.
Google's Gemma 4 scored 80% on a coding benchmark, up from 29% for Gemma 3 — nearly a 3x improvement in one generation.
On a math benchmark, Google's Gemma 4 scored 89% versus Gemma 3's 20% — more than a four-fold jump.
Chapter 15 · 34:20
Google just made video AI free — the same week OpenAI reportedly shut down Sora. Coincidence?
While the world debates Claude vs. GPT, Google has quietly built the best model in every single vertical: deep research, video generation, image generation, on-device inference. Then they opened Google Vids for free — right after OpenAI reportedly shut down Sora. OpenAI is getting hit from all sides.
Intel's stock rose approximately 30–31% over the past thirty days as its chip manufacturing partnerships with Google and xAI expanded.
Chapter 16 · 38:15
Luca had the right Intel thesis and sold too early. The stock is now up 30% in 30 days.
Intel is up 30% in 30 days as xAI, Google, and others sign chip manufacturing deals. Luca had the right thesis — Intel's existing facilities let partners skip the permitting nightmare — but got cold feet and sold. The lesson: let your winners ride.
Chapter 17 · 42:25
ZAI's GLM 5.1 is being called the first open-source model that actually feels like Opus. Is the endgame here?
ZAI's GLM 5.1 is being called the first open-source model that genuinely approaches Opus-level capability. If true, the endgame Luca predicted — where AI intelligence becomes a commodity and users stop migrating every six months — may be arriving faster than expected.
Chapter 18 · 44:00
Six spots. Malta. Founder retreat. You have to be building with AI to get in.
The Godmode Pod team is organizing a founder retreat in Malta with only six spots. You need to be building with AI to qualify, and you have to talk to the hosts first. More details dropping soon — stay subscribed if you want in.
No indexed bits in this chapter.
Mythos, deployed quietly under Project Glasswing to 20 strategic partners including Apple, Google, and Microsoft, found thousands of high-severity vulnerabilities — including a 27-year-old flaw in OpenBSD. If adversaries get a similar model, the banking system and critical infrastructure could be catastrophically exposed.
While the world debates Claude vs. GPT, Google has quietly built the best model in every single vertical: deep research, video generation, image generation, on-device inference. Then they opened Google Vids for free — right after OpenAI reportedly shut down Sora. OpenAI is getting hit from all sides.
This episode
Factual claims made this episode, and whether a source was named.
Anthropic's Mythos model scored 78 on a coding benchmark where GPT-5.4 scored 58.
Mythos preview found thousands of high-severity vulnerabilities across every major operating system and web browser, including a 27-year-old bug in OpenBSD.
Anthropic accidentally left a draft blog post about Mythos in an unsecured, publicly searchable data store, which Fortune discovered on March 26.
Anthropic released Mythos preview to 20 strategic partners including AWS, Apple, Broadcom, Cisco, Google, and Microsoft on April 7.
Mythos is approximately 50% more powerful than Anthropic's previous flagship Opus model and scores roughly 25 percentage points higher on benchmarks.
Jensen Huang stated that a $250,000-per-year software engineer who is not spending at least $500,000 in AI API token credits has something seriously wrong with them.
Meta has an internal leaderboard rewarding employees who spend the most on AI LLM tokens.
Anthropic has overtaken OpenAI in revenue, hitting a $30 billion annualized run rate.
OpenAI projected $100 billion in advertising revenue by 2030.
OpenAI acquired TBPN for approximately $200 million, roughly one year after TBPN launched.
Google's Gemma 4 scored 89% on a math benchmark, up from Gemma 3's 20%, and 80% on coding, up from 29%.
Google opened its Vids video generation tool (powered by Veo 3) for free to all users, removing the previous requirement of a Google AI subscription.
Intel's stock rose approximately 30–31% over the previous thirty days amid new AI chip manufacturing partnerships.
Anthropic trains its models approximately four times more efficiently than OpenAI, requiring only a quarter of the compute for equivalent capability.
Anthropic's Opus model costs approximately two to three times more per token than Sonnet.
This episode
NVIDIA CEO cited for the claim that a $250K engineer not spending $500K on AI tokens is underperforming.
OpenAI CEO mentioned in the context of the TBPN acquisition and OpenAI's media strategy.
Central company of the episode — discussed for Mythos launch, OAuth cancellation, managed agents, revenue growth, and compute constraints.
Praised for its diversified AI strategy across Gemma, Gemini, Vids, and image models, and highlighted as potentially the quiet winner of the AI race.
Discussed for its ad revenue ambitions, TBPN acquisition, ChatGPT distribution strategy, and being overtaken by Anthropic in revenue.
Discussed for its surging stock (+30% in 30 days) and new chip manufacturing partnerships with Google and xAI/SpaceX/Tesla.
Silicon Valley tech podcast acquired by OpenAI for approximately $200 million, just one year after its launch.
Referenced for having an internal leaderboard that rewards employees who spend the most on AI API tokens.
Referenced through CEO Jensen Huang's claim that engineers should spend at least $500K in AI API token credits per year.
Named as one of 20 strategic partners given early access to Anthropic's Mythos model under Project Glasswing.
Referenced as an example of a breakthrough architectural approach — distillation and novel training — that enabled a step-change in model efficiency.
Anthropic's leaked next-generation model, scoring 78 on a coding benchmark vs GPT-5.4's 58, and used in Project Glasswing to find cybersecurity vulnerabilities.
Open-source framework allowing users to run Claude-powered bots via Telegram and WhatsApp; effectively disabled when Anthropic cancelled OAuth access.
Google's open-weight model that runs offline on phones, jumping from 29% to 80% on coding benchmarks and 20% to 89% on math in one generation.
Discussed as OpenAI's consumer-facing product being used as a distribution play, with free users likely to be monetized through advertising.
Google's video generation tool (powered by Veo 3) opened for free to all users, directly competing with OpenAI's Sora.
Chinese open-source model described as the first to genuinely approach Anthropic's Opus in coding capability.
Hardened open-source operating system in which Mythos discovered a 27-year-old security vulnerability under Project Glasswing.
Stats
We use essential and analytics cookies to run Vuci. To understand how the site is used: Privacy Policy.
Install Vuci on your phone
Add it to your home screen for a faster, app-like experience.
Install Vuci on your phone
Tap the Share button, then “Add to Home Screen”.
A new version is available
Reload to get the latest Vuci.