Sam Altman on Why OpenAI Kept Token Pricing
OpenAI considered dropping token-based billing entirely for a flat quote per prompt, but enterprise customers pushed back because they demanded an auditable, quantifiable unit.
Theme
Pricing: what the top podcasts are saying.
How operators set, test, and raise prices. 27 write-ups from 16 shows so far, the newest from November 2026.
27 write-ups16 shows
The short version
Shows agree that algorithmic pricing is replacing fixed tags by tracking consumer behavior and pooling competitor data. Concurrently, AI providers maintain strict token billing for enterprise buyers, while software founders test extreme price tiers to signal trust and replace human services.
Top talking points
Fixed prices compete with algorithmic surveillance
Fixed public tags historically function as a public good that keeps costs lower for shoppers. Meanwhile, food delivery platforms track device battery life to alter menu discounts, and reservation systems treat restaurant tables like expiring options contracts.
- Why Quaker Price Tags Beat Algorithmic Pricing Odd Lots · Nov 1
- Lindsay Owens on Algorithmic Pricing and Device Telemetry Odd Lots · Nov 1
- Nick Kokonas on Dynamic Pricing for Time-Slotted Businesses My First Million · Nov 1
Token pricing reflects underlying compute constraints
OpenAI maintains token billing because enterprise buyers demand an auditable unit for operational budgets. API providers scale these token rates based on hardware memory bottlenecks, while builders like Claire Vo output typed data primitives to eliminate text generation costs.
- Sam Altman on Why OpenAI Kept Token Pricing Cheeky Pint · Nov 1
- LLM API Costs: Context Length Reveals Memory Bottlenecks Dwarkesh Podcast · Apr 26
- Claire Vo: Why Decision Models Cost 4 Cents per Million Tokens How I AI · Oct 4
Extreme fees establish immediate product trust
Founders charge up to $10,000 a month by positioning products directly against expensive human services. Venture capitalists like Anish Acharya note a premium tier establishes immediate quality for skeptical buyers and measures peak willingness to pay.
- Why Mentava Charges $500 a Month to Teach Toddlers to Read The Tim Ferriss Show · Sep 13
- Price Consumer Software Like a Birkin Bag: Anish Acharya on Ambition Lenny's Podcast · Sep 6
Automated agents require static pricing data
Frontier models hallucinate maximum market rates when web crawlers encounter interactive JavaScript pricing sliders. Startups manage these AI expenses by building multi-model routing systems that dynamically send requests to the cheapest capable endpoint.
- Why AI Models Kept Calling Resend the Most Expensive Email Service The Changelog · Oct 25
- Lovable Hit $500M: Their AI Mistake-Learning Loop & Token Pricing All-In Podcast · Jul 19
High win rates indicate underpriced software
Enterprise sales win rates above 35% signal that a company leaves money on the table. Sales teams wait to negotiate exact pricing terms until a pilot establishes internal momentum.
- Why High Win Rates Mean Your Enterprise Price Is Too Low Lenny's Podcast · Aug 23
- How ElevenLabs Scaled Past $450M ARR at an $11B Valuation Cheeky Pint · Apr 19
Most interesting insights
Groundwork Collaborative uncovered a DoorDash patent featuring a 16-factor agitation index that tracks scrolling speed and device battery life to alter menu ranks and discounts.
From Lindsay Owens on Algorithmic Pricing and Device Telemetry, Odd Lots · Nov 1
75% of items had different prices for different shoppers buying from the exact same store at the exact same second.
From Instacart: Secret Shoppers Found 75% of Groceries Varied in Price, Odd Lots · Nov 1
High software pricing acts as an immediate trust filter for skeptical parents who assume cheap educational apps are low-quality scams.
From Why Mentava Charges $500 a Month to Teach Toddlers to Read, The Tim Ferriss Show · Sep 13
Stripe currently spends about 1% of its total operating expenses on tokens, and John Collison expects that figure to climb to 5%, 10%, and eventually 20%.
From Sam Altman on Why OpenAI Kept Token Pricing, Cheeky Pint · Nov 1
From our research reports
The Monthly Read: October 2026
Hardware makers and enterprise buyers are using open-weight models to erode the pricing power of closed API providers
Across 5 shows, commentators agreed that enterprise workloads are tilting toward open-weight models to gain internal auditing, weight modification, and lower operating costs compared to closed APIs. Both chip suppliers and enterprise buyers view open software as a structural tool to redirect software margins into owned hardware and local infrastructure.
Where they split: Speakers clashed over whether foreign open-weight foundation models offer an economic windfall or an unacceptable governance liability. While some investors argue these releases function as a direct research subsidy to Western software builders, others highlight that corporate buyers aggressively push back against deploying foreign models due to regulatory and security risks.
People quoted on pricing
Latest write-ups
Lindsay Owens on Algorithmic Pricing and Device Telemetry
Personalized pricing is rebranded first-degree price discrimination, using real-time surveillance to strip away consumer surplus.
Instacart: Secret Shoppers Found 75% of Groceries Varied in Price
Groundwork Collaborative, Consumer Reports, and More Perfect Union recruited 400 secret shoppers across four US markets to audit Instacart orders simultaneously.
Why Quaker Price Tags Beat Algorithmic Pricing
The fixed price tag is only about 150 years old, popularized in the late 1870s by Philadelphia department store founder John Wanamaker.
Nick Kokonas on Dynamic Pricing for Time-Slotted Businesses
Nick Kokonas generated over $100M trading options by exploiting latency between Chicago and New York before applying those same principles to hospitality.
Algorithmic Price-Fixing Is Quietly Killing Competition
RealPage enabled competing landlords to share sensitive market data and coordinate prices, driving apartment rents up to 25% above competitive market levels.
Why AI Models Kept Calling Resend the Most Expensive Email Service
Resend discovered that Claude and ChatGPT were telling prospective buyers the platform was the most expensive email service on the market, despite offering competitive rates.
Brookfield Finds $300M in EBITDA Through Industrial AI
Brookfield PE CEO Anuj Ranjan oversees roughly 1,200 active AI deployments across the firm's global portfolio, generating about $300 million in recurring EBITDA.
How the Diamond Industry Built Language for Price Discrimination
De Beers and the Gemological Institute of America established the Four Cs to create price discrimination for differences invisible to the human eye.
Why Carve-Out TSAs Fall Apart After the LOI
Buyers and sellers regularly face a 15-month duration mismatch, with buyers expecting 18 months of operational runway while sellers plan to exit after 90 days.
Nancy Zimmerman on Pricing Securities vs. Investing
Nancy Zimmerman co-founded Bracebridge Capital in 1994 alongside Gabe Sunshine, backed early by Yale endowment head David Swensen.
Claire Vo: Why Decision Models Cost 4 Cents per Million Tokens
Standard frontier models generate unstructured text strings, creating high latency, unpredictable output parsing, and steep output token costs.