“while prices for some models might have come down, newer “reasoning” models burn way more tokens, which means the cost of inference has, somehow, gotten higher over time.”
Our reading, the sentence that was checked:
Although prices for some models might have come down, newer reasoning models burn far more tokens, so the cost of inference has gotten higher over time.
Assessment: Mixed Evidence as of
Where and when Ed Zitron wrote it
Where's Your Ed At, AI's Economics Don't Make Sense, paragraph 22.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment wyea-ais-economics-dont-make-sense-b79382-a2
We recorded 31 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 2026-04-28. 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
- while prices for some models might have come down (qualifier, stated by the speaker)
From the speaker
- prices for some models might have come down
- newer “reasoning” models burn way more tokens
- which means the cost of inference has, somehow, gotten higher over time
Our interpretation
- The ordinary reading concerns the total cost of performing work with newer reasoning models, rather than the price of one token.
- The claim combines a mechanism, increased token use, with a broad historical cost conclusion.
- Comparable-task evidence must distinguish running increasingly capable frontier models from achieving a fixed level of performance.
- The historical endpoint is publication. No starting date is supplied, so the lower bound remains unspecified.
- API expenditure can measure inference expense paid by an application, but does not establish model providers' internal costs.
Counts as supported if: Comparable-task measurements show that increased reasoning-token use raised total inference expenditure across newer frontier models despite lower prices elsewhere.
Counts as contradicted if: Measurements including reasoning-token consumption show falling expenditure to achieve comparable performance. If frontier expenditure rises while fixed-performance costs fall, the compound claim warrants a mixed assessment.
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 April 2024–November 2025 study supports the narrower observation that expenditure on some frontier workloads rose: overall benchmarking costs stayed flat or increased, while costs at a given performance level fell [e8]. Its authors suggest larger models and longer reasoning traces as an explanation; this does not establish reasoning-token growth as the cause [e8]. The broad conclusion that comparable inference became more expensive does not hold in these measurements: the study reports falling fixed-performance costs, and the coding analysis published September 17, 2025 also finds falling expenditure at fixed benchmark scores [e8, e9].
Limitations
- The studies do not establish an industry-wide trend across production workloads or providers' internal inference costs [e8, e9].
- The frontier-cost study does not isolate reasoning-token growth from model size or demand for higher performance [e8].
Evidence (3 items)
Supports the claim: arXiv — Gundlach, Lynch, Mertens and Thompson,
reasoning traces.
Why it matters here: The study's April 2024–November 2025 data show increasing frontier expenditure alongside longer reasoning and larger models. This supports the cost-escalation mechanism for frontier workloads, although its causal attribution is observational.
- Publisher
- arXiv — Gundlach, Lynch, Mertens and Thompson
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
Goes against the claim: Håvard Tveit Ihle / LessWrong,
a halving time
Why it matters here: The analysis finds falling total inference expenditure at fixed benchmark scores, with halving times measured in months. This contradicts an unqualified rise in the cost of comparable performance, while remaining compatible with more spending on the frontier.
- Publisher
- Håvard Tveit Ihle / LessWrong
- Kind of source
- Independent measurement
- Describes events of
- Public from
Context only: Kilo Code,
test-time scaling
Why it matters here: Zitron's linked source describes rising application spending from more tokens, larger contexts and reasoning. It does not provide a controlled historical comparison holding completed work and quality constant.
- Publisher
- Kilo Code
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Where it was said (1)
Said once in the sources this register covers.
: Where's Your Ed At, AI's Economics Don't Make Sense
while prices for some models might have come down, newer “reasoning” models burn way more tokens, which means the cost of inference has, somehow, gotten higher over time.
- Where
- paragraph 22
- Published
- 2026-04-28
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment wyea-ais-economics-dont-make-sense-b79382-a2: 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-ais-economics-dont-make-sense-b79382-a2.