AI token costs have crashed 850x since 2020 — what cost $60 per million tokens now costs 6 cents — and one host is burning $12,500 worth of tokens a month on a $200 subscription.
God Mode Podcast
AI token costs have crashed 850x since 2020 — what cost $60 per million tokens now costs 6 cents — and one host is burning $12,500 worth of tokens a month on a $200 subscription.
TL;DR
Three AI practitioners dissect the week's biggest stories: Anthropic's rejection of a Pentagon surveillance contract (and OpenAI's same-day swoop to claim it), the deep ties between effective altruism and Anthropic's decision-making, and a jaw-dropping chart showing AI token costs have fallen 850x since 2020. They also debate whether running AI locally on a Mac Studio makes sense (verdict: not yet), the collapse of AI avatar startup Icon.com versus Higgs Field's explosive growth, and GPT-5.4's surprising computer-use benchmark. Key takeaway: plan your AI system architecture before you build it, or you'll rebuild it from scratch.
Three AI practitioners debate Anthropic's Pentagon contract rejection, OpenAI's government deal, effective altruism's grip on Anthropic, the 850x crash in token costs, and the real-world tradeoffs of running AI locally versus in the cloud.
Chapter 3 · 03:05
Anthropic missed a 5PM Pentagon deadline. By 11PM, Sam Altman had tweeted the deal.
The Pentagon gave Anthropic a hard 5PM deadline to accept unrestricted military use of Claude. Dario was in meetings and missed it. By 11PM, Sam Altman had tweeted a done deal with OpenAI. What looked like a principled stance was also, at least partly, a scheduling failure — but the PR outcome for Anthropic was paradoxically positive.
Claude hit the number one spot in the App Store in the same week Anthropic turned down the Pentagon contract, suggesting the ethical stance was a net PR win.
In the same period Claude hit the top of the App Store, ChatGPT downloads reportedly dropped by around 250%, indicating a meaningful consumer sentiment shift.
After Anthropic missed the Department of Defense's deadline, OpenAI signed a $200 million per year government contract the same evening.
Chapter 4 · 09:20
Anthropic's CEO, COO, investors, and major donors are all one EA social network.
Anthropic's founders, funders, and key executives are almost all connected through the EA movement. Dario's college roommate co-founded Open Philanthropy, married Dario's sister Daniela, and joined Anthropic in 2025. SBF's $500M investment used FTX customer funds. EA isn't just Anthropic's philosophy — it's its entire social graph.
Sam Bankman-Fried invested $500M of FTX customer funds into Anthropic, receiving a 13.5% stake now diluted to 8%. At Anthropic's current $380B valuation, that stake is worth roughly $30 billion — enough to repay every FTX victim three times over. The best VC call in history was made illegally.
Sam Bankman-Fried invested $500M in Anthropic for a 13.5% stake (diluted to 8%) that would be worth ~$30 billion at today's $380B valuation — enough to repay FTX victims three times over.
Chapter 5 · 21:10
I pay $200/month for AI. I'm actually using $12,500 in tokens. The math doesn't work.
Anthropic's most recent valuation stands at approximately $380 billion, always roughly half of OpenAI's valuation despite a much smaller user base.
Ben pays $200/month for Claude Max but burns roughly 500 million tokens a month — equivalent to $12,500 at API rates. AI providers priced subscriptions like a normal buffet, but developers are now taking entire restaurants' worth of food home. The OAuth crackdown from Anthropic and Google is the inevitable response.
Chapter 6 · 22:20
Ben pays $200/month for a Claude Max subscription but consumes roughly 500 million tokens — equivalent to $12,500 at API pricing — demonstrating how broken subscription economics are.
Ben pays $200/month for a Claude Max subscription but consumes roughly 500 million tokens — equivalent to $12,500 at API pricing — demonstrating how broken subscription economics are.
Chapter 7 · 26:15
AI tokens cost 850 times less than they did in 2020. This changes everything.
AI token costs have fallen 850 times since 2020. What cost $60 per million tokens with GPT-3 now costs 6 cents with open-source Llama 4. It's not just cheaper — the models are dramatically smarter. Every industry bottleneck unlocked by intelligence is now also being unlocked by near-zero cost.
OpenAI's frontier model price per million tokens dropped from $30 in Q1 2023 to $0.50 today; Anthropic went from $15 to $1, Google from $7 to $0.07.
Chapter 8 · 28:50
Every time OpenAI improves, the open-source models that compete with it get better too.
Open-source models are mostly distilled from the latest closed-source frontier models. So every improvement OpenAI or Anthropic makes also improves the open-source alternatives that compete with them. The companies funding frontier research are inadvertently subsidising their own open-source rivals.
Chapter 9 · 32:30
Claude Code already accounts for 4% of all GitHub commits and is projected to reach 20% by year-end — a remarkable rise for a single AI coding tool.
Claude Code already accounts for 4% of all GitHub commits and is projected to reach 20% by year-end — a remarkable rise for a single AI coding tool.
Chapter 10 · 34:50
Claude Code's framework just became the most-starred GitHub repo ever — bigger than React.
Claude Code's open-source framework has surpassed React as the most-starred GitHub repo of all time. Luca reads this as a signal that the winning consumer AI product won't come from a big lab — it'll be a closed, polished version of OpenClaw: model-agnostic, self-hosted, and shipped out of the box.
Chapter 11 · 36:30
What if the Jony Ive AI device just shipped with Claude Code pre-installed? That's the killer app.
Jony Ive and Sam Altman are reportedly building an AI device together. Luca's provocative take: the killer device isn't some futuristic gadget — it's one that ships with a secure, pre-configured version of Claude Code's open-source framework, letting anyone run personalised AI agents out of the box without cloud dependency.
Chapter 13 · 42:30
Mac Studio fans: 34 seconds for 400 words locally vs 7 seconds in the cloud. Cloud wins.
Running a 32B quantised Llama model on a Mac Mini takes 34 seconds to write 400 words. Claude Sonnet in the cloud does it in 7 seconds. And the local model is meaningfully dumber. Unless you've fully systematised your workflow around dedicated hardware, local AI inference is slower, weaker, and barely more private.
Writing a 400-word section takes 34 seconds on a Mac Mini running a quantised 32B Llama model, versus 7 seconds using Claude Sonnet in the cloud — making local inference impractical for fast workflows.
Chapter 14 · 47:00
One AI startup spent $12M on its domain and went bankrupt. The other is heading for $1B ARR.
Icon.com raised big money from Peter Thiel and OpenAI, spent $12M on a domain, launched a chaotic landing page, and is apparently bankrupt. Higgs Field took a disciplined approach to AI video generation, grew from 15M users to 4.5M daily videos, and is on track for $1B ARR. Same AI wave, completely different execution.
AI avatar startup Icon.com spent $12 million on its domain name, was backed by Peter Thiel and OpenAI, and has since apparently gone bankrupt — a cautionary tale of mismanagement.
AI video platform Higgs Field grew its daily video generation from 15 million to 4.5 million per day and is on track for $1 billion ARR, contrasting sharply with rival Icon.com's collapse.
Chapter 15 · 53:10
Cluely's founder Roy publicly claimed $7 million ARR to TechCrunch when the company was actually doing closer to $4–5 million, issuing a public retraction describing it as 'the only blatantly dishonest thing I've said online.'
Cluely's founder Roy publicly claimed $7 million ARR to TechCrunch when the company was actually doing closer to $4–5 million, issuing a public retraction describing it as 'the only blatantly dishonest thing I've said online.'
Chapter 16 · 57:30
GPT-5.4 just clicked its way past the human baseline on real computer use benchmarks.
GPT-5.4 jumped from 47% to 75% on OSWorld, a benchmark testing real desktop navigation, in a single release — crossing the human baseline. It apparently clicks around the internet at human speed. Previous agent modes looked impressive but were painfully slow; this model may be the first where computer use is genuinely practical.
GPT-5.4 jumped from 47% to 75% on the OSWorld computer-use benchmark in a single release, surpassing the human baseline — a striking leap attributed to combining a coding model with a general model.
Chapter 17 · 1:04:00
SanDisk went 20x in a year. AI is a memory business — and most people missed it.
SanDisk stock rose from $30 to $600 — a 20x gain in a single year — because AI infrastructure is fundamentally a memory business. Samsung is up 250% for the same reason. These are the obvious-in-hindsight picks that everyone will claim they saw coming once the charts go vertical.
Chapter 18 · 1:06:50
SanDisk's stock rose from a low of around $30 to $600 — a roughly 20x gain in a single year — driven by surging AI demand for memory storage.
SanDisk's stock rose from a low of around $30 to $600 — a roughly 20x gain in a single year — driven by surging AI demand for memory storage.
Chapter 19 · 1:08:50
AI made you more productive. It made your competitors more productive too. Net result: harder.
AI genuinely makes individuals more productive. But it has also turbo-charged every competitor. Landing pages are better, content is cheaper, and growth channels are flooded. The net result in SaaS: more capability, more competition, similar difficulty. The humans are still winning on outreach and relationship-building.
Chapter 20 · 1:16:10
Zoom out before you build with AI — or you'll spend two weeks building something you delete.
The most common AI workflow mistake is skipping the architecture phase. Rick rebuilt his first vibe-coded project from scratch in 48 hours after two weeks of messy iteration because he hadn't planned the full system. Map every step, every agent connection, and every tool integration before writing a single prompt.
The cost of one million AI tokens fell from $60 with GPT-3 DaVinci to 6 cents with open-source Llama 4 — an 850-fold decrease in roughly five years.
No indexed bits in this chapter.
AI token costs have fallen 850 times since 2020. What cost $60 per million tokens with GPT-3 now costs 6 cents with open-source Llama 4. It's not just cheaper — the models are dramatically smarter. Every industry bottleneck unlocked by intelligence is now also being unlocked by near-zero cost.
Ben pays $200/month for Claude Max but burns roughly 500 million tokens a month — equivalent to $12,500 at API rates. AI providers priced subscriptions like a normal buffet, but developers are now taking entire restaurants' worth of food home. The OAuth crackdown from Anthropic and Google is the inevitable response.
The Pentagon gave Anthropic a hard 5PM deadline to accept unrestricted military use of Claude. Dario was in meetings and missed it. By 11PM, Sam Altman had tweeted a done deal with OpenAI. What looked like a principled stance was also, at least partly, a scheduling failure — but the PR outcome for Anthropic was paradoxically positive.
This episode
Factual claims made this episode, and whether a source was named.
The Pentagon's deadline for Anthropic was 5PM on February 24, 2026, set by Secretary Pete Hegseth, requiring unrestricted use of Claude for all legal purposes.
Sam Altman announced via tweet by 11PM on February 27 that OpenAI had struck a deal with the Pentagon, the same evening Anthropic missed its deadline.
The Pentagon government contract signed with OpenAI is worth $200 million per year.
Claude was the number one app in the App Store in the week Anthropic refused the Pentagon contract.
ChatGPT downloads dropped by approximately 250% in the same period Claude surged to #1 in the App Store.
Sam Bankman-Fried invested $500 million in Anthropic in April 2022 using FTX customer funds, receiving a 13.5% equity stake that has since diluted to approximately 8%.
SBF's 8% diluted Anthropic stake would be worth approximately $30 billion at Anthropic's current valuation — enough to repay FTX creditors roughly three times over.
Anthropic is currently valued at approximately $380 billion and is consistently valued at roughly half of OpenAI's valuation.
AI token prices have fallen 850 times since 2020, from $60 per million tokens for GPT-3 DaVinci to 6 cents for open-source Llama 4.
OpenAI's frontier model token price dropped from $30 per million in Q1 2023 to $0.50 today; Anthropic's from $15 to $1; Google's from $7 to $0.07.
Ben pays $200/month for Claude Max but consumes roughly 500 million tokens per month — equivalent to approximately $12,500 at API pricing.
Claude Code accounts for approximately 4% of all GitHub commits and is projected to reach 20% by end of 2025.
Icon.com spent $12 million on its domain name, was funded by Peter Thiel and OpenAI, and has apparently gone bankrupt.
Higgs Field grew from 15 million users to generating 4.5 million videos per day and is on track for $1 billion in ARR.
Cluely founder Roy publicly told TechCrunch the company was doing $7 million ARR when it was actually around $4–5 million, later issuing a public retraction.
GPT-5.4 scored 75% on the OSWorld computer-use benchmark, up from 47%, surpassing the human baseline.
A Mac Mini running a quantised 32-billion-parameter Llama model takes 34 seconds to write a 400-word section, while Claude Sonnet in the cloud completes the same task in 7 seconds.
SanDisk's stock rose from a low of approximately $30 to $600 — roughly a 20x gain — within the past year, driven by AI memory demand.
Samsung stock is up approximately 250% in the past year due to AI chip and RAM manufacturing demand.
Copper is up approximately 20% in the past year and is at an all-time high, breaking into price discovery mode.
This episode
Anthropic CEO discussed for missing the Pentagon deadline and for his EA philosophy and views on AI governance.
OpenAI CEO who tweeted the Pentagon deal by 11PM on the same day Anthropic missed its deadline; described as fully capitalist in strategy.
Discussed for investing $500M of FTX customer funds in Anthropic at a 13.5% stake that would now be worth ~$30B — and as a cautionary EA case study.
Anthropic co-founder and COO, sister of Dario Amodei; her husband co-founded Open Philanthropy, deepening EA ties to Anthropic.
Former Apple designer reportedly partnering with Sam Altman on an AI device; speculated by hosts to potentially ship with Claude Code pre-installed.
Facebook co-founder and Open Philanthropy funder; a major early backer of Anthropic and central node in the EA-Anthropic network.
US Secretary of Defense who gave Anthropic CEO Dario Amodei a hard 5PM deadline to accept unrestricted military AI use, triggering the contract drama.
Central to the episode: Anthropic's Pentagon contract refusal, its EA-aligned culture, Claude's market rise, and its pricing strategy are the dominant threads.
Discussed as the beneficiary of Anthropic's Pentagon contract rejection and as a capitalist-first counterpart to Anthropic's EA-driven model.
AI video generation startup growing explosively to 4.5M videos/day and on track for $1B ARR — contrasted with Icon.com's collapse.
AI meeting/cheating app startup that misrepresented its ARR to TechCrunch, has pivoted multiple times, and is seen as emblematic of the AI hype cycle deflating.
AI avatar startup funded by Peter Thiel and OpenAI that spent $12M on its domain and reportedly went bankrupt, cited as a cautionary mismanagement tale.
Sam Bankman-Fried's collapsed crypto exchange; its $500M Anthropic investment using customer funds is central to the EA-Anthropic origin story.
EA's largest philanthropic fund and one of Anthropic's earliest major backers; co-founded by Dustin Moskovitz and Dario Amodei's college roommate.
Chip and RAM manufacturer up 250% in the past year, discussed as a key AI infrastructure beneficiary due to chip manufacturing demand.
Memory storage company whose stock rose roughly 20x in a year, driven by AI infrastructure demand — cited as a missed-but-obvious investment.
Anthropic's AI model, praised as best-in-class for developers; reached #1 in the App Store after the Pentagon story broke.
Anthropic's agentic coding tool; accounts for 4% of GitHub commits and is projected to reach 20%; also discussed as the most-starred GitHub repo framework.
Discussed as a 'sleeping giant' offering significantly cheaper token pricing than OpenAI or Anthropic, quietly gaining market share via free tiers.
Vibe-coding tool discussed as an early-mover in AI-assisted app development whose improvements largely derive from underlying model advances at Anthropic.
Stats
We scan show notes for social handles, websites and apps. Nothing matched on this episode.
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.