“A coding model relies heavily on chain-of-thought reasoning, which means that despite the cost of tokens coming down (which does not mean the price of providing them has decreased, it’s a marketing move), models are using far, far more tokens, increasing costs across the board.”
Our reading, the sentence that was checked:
Coding models rely heavily on chain-of-thought reasoning and so use far more tokens, which raises costs across the board even though token prices have fallen; lower token prices do not mean the cost of providing tokens has fallen, and are a marketing move.
Assessment: Mixed Evidence as of
Where and when Ed Zitron wrote it
Where's Your Ed At, The Subprime AI Crisis Is Here, paragraph 67.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment wyea-the-subprime-ai-crisis-is-here-e5bcdd-a1
We recorded 43 claims from this post. See every claim from this post.
The wording above is the version this site observed on 2026-10-28. The source may have been edited after 2026-03-31. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Causal claim
- Deadline, as the speaker wrote it
- None stated. This site does not add one.
- Deadline state when assessed
- No horizon
- Deadline state on 2026-10-29
- No horizon (the release’s reference date)
Conditions
- coding models (scope, stated by the speaker)
From the speaker
- A coding model relies heavily on chain-of-thought reasoning
- models are using far, far more tokens
- increasing costs across the board
- the cost of tokens coming down
- it’s a marketing move
Our interpretation
- The plain reading is that reasoning materially increases coding models’ token consumption and makes their use more costly despite falling token prices.
- Assess the reliance on reasoning separately from the broader claims about rising costs and the reason for price reductions.
- Compare equivalent coding tasks and account for task success, model choice and reasoning settings. Distinguish customers’ API bills from providers’ serving costs.
- The statement describes a mechanism operating by publication, not a forecast.
Counts as supported if: Coding-task comparisons show substantial additional reasoning tokens and higher total costs despite lower token prices; provider evidence establishes that price cuts reflect marketing rather than lower serving costs. Evidence can support individual components without establishing the whole claim.
Counts as contradicted if: Comparable coding-task results show newer models achieving equivalent or better results with fewer tokens or lower total costs, contradicting the across-the-board generalization. Disclosed serving efficiencies or pricing decisions could separately contradict the marketing explanation.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Does the evidence support the cause?
Outcome: Mixed
Part of the claim holds and part does not, and the record says which part is which.
Scored on a plain reading. The speaker named no measure for this claim, so it was judged on the reading an informed reader would take, stated below under our interpretation. Both reviews had to find that reading fair.
- Evidence as of
- Deadline state when assessed
- No horizon
- Assessment recorded
- Policy version
- 2026-10-28.7
Rationale
The reliance-on-reasoning component has direct support from Cursor. The broader token-growth claim has a documented exception: Anthropic reports better coding-benchmark performance with 48% fewer tokens. That challenges the across-the-board wording. The evidence does not settle aggregate serving costs or the claim that falling prices are a marketing move.
Limitations
- The decisive coding evidence is company-reported.
- A counterexample does not disprove a market-wide average trend.
- The mixed assessment concerns identifiable components; it does not establish the claimed financial mechanism.
Evidence (3 items)
Supports the claim: Cursor,
critical to the model’s intelligence.
Why it matters here: Cursor identifies reasoning stages as important to its coding model. This supports the reliance-on-reasoning component, without establishing industry-wide token growth or costs.
- Publisher
- Cursor
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Goes against the claim: Anthropic,
using 48% fewer tokens.
Why it matters here: This coding comparison reports better performance with fewer tokens. It contradicts an across-the-board increase in token requirements, although it does not measure aggregate provider spending.
- Publisher
- Anthropic
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Context only: arXiv,
higher inference latency and memory consumption.
Why it matters here: The paper describes the resource burden of long reasoning traces. Its mathematics results and resource measures do not establish coding-service bills or providers’ financial costs.
- Publisher
- arXiv
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
What would settle it
Matched coding-task cost histories and provider serving-cost disclosures, together with records explaining price reductions.
Where it was said (1)
Said once in the sources this register covers.
: Where's Your Ed At, The Subprime AI Crisis Is Here
A coding model relies heavily on chain-of-thought reasoning, which means that despite the cost of tokens coming down (which does not mean the price of providing them has decreased, it’s a marketing move), models are using far, far more tokens, increasing costs across the board.
- Where
- paragraph 67
- Published
- 2026-03-31
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment wyea-the-subprime-ai-crisis-is-here-e5bcdd-a1: Current. Outcome: Mixed. Evidence as of 2026-10-28. Public date not recorded.
Recorded events
Assessment publishedBy this site
First assessment published after both reviews and the final-text check passed.
About assessment wyea-the-subprime-ai-crisis-is-here-e5bcdd-a1.