Model

Respan served OpenAI's GPT-6.2 Sol at $0 for 24 hours

The plumbing for the giveaway was finished two weeks before OpenAI shipped the model. The money at risk is a $5 million seed betting free tokens turn into paid observability accounts.

Richard TangRichard Tangthree-minute read
Share
Sam Altman seated onstage at TechCrunch Disrupt SF 2017, gesturing with both hands as he speaks
Sam Altman onstage in 2017. His OpenAI held GPT-6.2 Sol's list price while Respan served the model at $0 for a day.

OpenAI$1.18T — OpenAI, private, latest valuation $1.18T shipped GPT-6.2 Sol on November 17 and left its list price unchanged 1. The next day Respan served the same model at $0 in and out for 24 hours: one API key, working in the coding tools Claude Code, Cursor and Codex, no card on file 23. Respan is a 17-person startup selling the gateway layer between developers and AI providers 34, and a free day on a brand-new model is how it buys distribution in a market where the incumbent's position has just been valued in the billions 5.

The promotion machinery shipped before the model did

Respan's November 4 changelog added free model promotions: call an eligible model with a ":free" suffix, and the response returns the remaining allowance; routing rules can target free models 3. A week later came a warning email at 80% of the allowance 3. The design caps the subsidy: free accounts hit a usage limit, and a $5 deposit unlocks more 6. A model launch is now a distribution event, if the gateway is wired for it.

OpenAI collected its list price either way

Respan makes its money on the hosted platform, on observability and evals, the logging and automated testing of what models return, not on the open-source gateway 7. Nothing published says who paid for the free tokens. If the bill was Respan's alone, every one was bought at Sol's rates, $2 in and $10 out per million, and given away 1.

Unless OpenAI quietly shared the cost, the one party sure to collect is OpenAI, whose list price held while somebody else paid it. The subsidy lands on Respan's books, and behind them a $5 million seed led by the venture firm Gradient Ventures, with Y Combinator participation 8.

The layer Stripe just priced in the billions

Forbes reported in August that Stripe had agreed to acquire OpenRouter, the incumbent router, for more than $8 billion in cash and stock, a deal neither company had then announced as closed 5. OpenRouter charges 5.5% on credit purchases 5. Respan's pitch is the mirror image: no fee, with bulk-rate discounts on tokens passed through 10.

Respan's example bill puts $5K of gpt-6.1-sol tokens about $1.2K under OpenRouter

No fee plus bulk rates is the pitch against OpenRouter's 5.5%

$0K$2K$4K$6KOpenRouter (+5.5% fee)$5.3KOpenAI list price$5KRespan (0% fee)$4.1K
Data
Value
OpenRouter (+5.5% fee)$5.3K
OpenAI list price$5K
Respan (0% fee)$4.1K
Respan's own homepage comparison for $5,000 of gpt-6.1-sol spend at list price; the company's numbers, not independently verified. US dollars in thousands.10

The number that decides it is conversion

Free signups are not revenue; revenue is the hosted platform 7. Respan itself ran on LiteLLM, an open-source gateway, from 2024 until this year's rewrite in Rust 7, and in March two malicious LiteLLM releases on PyPI stole the API keys of teams that installed them 9. That breach reset this market, and Respan has been marketing itself as the post-breach alternative since 9.

What no one outside the company can see is whether a free-day account ever becomes a paying one; the giveaway drew no coverage beyond developer posts 26. The subsidy was capped by design, and cheap next to what this layer now fetches 5. The watch-items are whether free-day accounts turn into paid observability revenue, and whether the free window returns at the next model launch.

Get Model by email, free

The AI labs, their models, and the business behind them. In our daily email, every weekday.

Reader comments

0 comments

    Model

    Google opens Gemini 4 Neon to all developers at $3 per million input tokens

    Google's fourth Gemini 4 model in five weeks pairs top-three benchmark scores with mid-tier pricing, resetting the margin math for AI software vendors in the very week Anthropic's record IPO slid back toward its $83 offer price.

    Model

    House report says the model behind SpaceX's $60 billion Cursor deal was built on China's Kimi K2.5

    SpaceX paid about $60 billion in stock for Cursor in August, and a House report now says its flagship model rests on Chinese open weights. The same day, rival Cognition entered talks at $75 billion, two months after Cursor's own backers led its $48 billion round.

    Model

    Mira Murati's Thinking Machines ships a 26-billion-parameter model that runs offline on a MacBook

    The 26-billion-parameter open-weights agent shipped hours after OpenAI said attackers had extracted GPT-6 Astra's hidden reasoning at a 54% success rate and asked Congress for new enforcement powers. Together the two moves frame the scarcity bet behind frontier token prices and the market's newest trillion-dollar listing.

    Model

    OpenAI says attackers copied GPT-6 Astra's hidden reasoning on 54% of tries; its fix cuts that to 0.2%

    OpenAI's November 20 disclosure puts a hard number on how copyable a frontier model was, and pairs the fix with a call for federal enforcement. The labs it names, Moonshot AI and DeepSeek among them, are where the money moves.

    Model

    Amazon's $188 Billion Anthropic Stake Swings $1.9 Billion for Every $1 of Anthropic Stock

    Two trading days turned Amazon's largest AI holding into a daily mark-to-market item, and Anthropic's second-session slide shows how fast the gain Amazon already booked can reverse.

    Model

    HeyGen sells Avatar 6 while Tavus's Griffin scores 3.83 against a human's 3.92 and still has no price

    HeyGen put Avatar 6 on general availability on November 17 with a public API and first place in New Measure's blind Avatar Arena; Tavus's Griffin, which Tavus says is the highest score ever recorded for an AI on NVIDIA's face-to-face benchmark, is still a preview with no price. Revenue accrues to HeyGen today because Avatar 6 is the model on sale; Tavus's best model is not.

    Subscribe to TheInference for free