Season 0 Join free
Models & labs · Thesis 87 · settles Jan 1 · 66 days

Flagship AI token price $3.90 or more, late December?

7 voices · 7 claims2 sides2 machines

The other numbers are hidden until you lock yours. The readers, the machines and the Call are waiting.

The two sides

If no

Cost per token falls faster than price

Expects compute costs to stay tight and token prices to hold firm even as usage scales.

If yes

The case against it

Expects efficiency gains and competition to keep pushing flagship token prices down below this mark.

How it settles

Yes if the seven day average of the flagship token price basket on deepstack.ltd market series, December 25 to 31, 2026, is 3.90 dollars per million tokens or higher. DeepStack public market series, flagship token price basket (daily, with history)

The exact question and the full rule

Will the flagship token price basket in DeepStack's public market series average at or above $3.90 per million tokens over the seven days from December 25, 2026 through December 31, 2026?

Take the daily value of the basket token-price-flagship in DeepStack's public market series (deepstack.ltd/api/v1/market-series.json) for each day from December 25, 2026 through December 31, 2026, and average the daily values. Each day's value is the median of the prices of the models that the OpenRouter aliases ~openai/gpt-sol-latest, ~anthropic/claude-opus-latest, ~google/gemini-pro-latest, ~x-ai/grok-latest, ~deepseek/deepseek-pro-latest, ~moonshotai/kimi-latest and ~z-ai/glm-latest point to, each blended as three input tokens to one output token in US dollars per million tokens, counted when at least four of the seven are recorded; each price is the first print the file holds for that day, recorded on or before the end of that day (UTC). The answer is yes if the mean is at or above $3.90 per million tokens; no if it is below $3.90 per million tokens.

If the source is late or silent: A day with no basket value is left out of the mean; if three or more of the seven days have none, the test is inconclusive. No other source replaces a missing day, and a print recorded after its day does not count. The test is read by January 3, 2027.

The case · 7 voices · 2 minutes
Cost per token falls faster than price2 people argue noThe case against it5 people argue yes
Stronger if no · 2 voices

Cost per token falls faster than price

Michael SpencerMay 28

As inference costs decline through hardware improvements, software optimization, or volume leverage, a vendor can either cut prices or expand margins. Anthropic is expanding margins: compute spend per revenue dollar falling from 0.71 to 0.56 while…

Stronger if yes · 5 voices

The case against it

DeepValue CapitalAug 21

While compute supply remains constrained, the price AI providers can charge per token is falling, squeezing the spread between input costs and revenue and potentially undermining the profitability case for AI infrastructure buildout.

The Call's side, the strongest voice against it and what would change it appear after you lock your number.

Open the dossierThe thesisLock your number first
The dossier · 5 minutes
The machines' reasons

2 models ran this one. Their numbers and reasons appear after you lock yours.

The thesis

Microsoft gross margin up 100 basis points by mid-2027?

Falling cost per served token outpaces token price declines, so hyperscaler-reported gross margins expand despite surging AI inference volume and depreciation from the capex buildout.

All 16 questions