“The cost of inference has gone up. Statements otherwise are purely false, and are the opinion of somebody who does not know what he's talking about.”
Our reading, the sentence that was checked:
The real cost of AI inference has risen, and claims that it is falling are false.
Assessment: Unresolved Evidence as of
Where and when Ed Zitron wrote it
Where's Your Ed At, How To Argue With An AI Booster, paragraph 236.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment wyea-how-to-argue-with-an-ai-booster-15df5f-a2
We recorded 35 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 2025-08-25. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Observation
- 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)
From the speaker
- The cost of inference has gone up
- Statements otherwise are purely false
- cost-per-tokens
- use more tokens to create an output
Our interpretation
- Read this as a claim about the general direction of actual inference spending per application workload, including reasoning tokens, as applications adopt newer models.
- A representative workload trend is needed to establish the general claim. Neither a token-price reduction nor one expensive application settles that trend.
- The surrounding discussion refers to growth over the last two years. Use approximately August 2023 through the publication date, with the exact anniversary as an operational boundary.
- This is an industry service-cost claim, not a claim about a named company's financial accounts; metric_scope is null.
Counts as supported if: Representative longitudinal application data show rising inference spending per comparable unit of work after accounting for token consumption, model choice, caching, quality, and workload changes.
Counts as contradicted if: Representative longitudinal application data show falling inference spending per comparable unit of work over the period, including reasoning tokens and changes in deployed models.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Was it true for the period described?
Outcome: Unresolved
The record was checked, and the evidence available by the cutoff does not settle it.
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
On July 4, 2025, Cursor reported that most users' costs had stayed fairly constant, while its hardest requests cost an order of magnitude more than simple ones [e1]. On April 14, 2025, OpenAI reported that GPT-4.1 was 26% less expensive than GPT-4o for median queries [e2]. Neither disclosure establishes rising or falling costs across representative, comparable application workloads during August 25, 2023–August 25, 2025 [e1, e2].
Limitations
- The disclosures do not provide representative longitudinal application bills linked to completed work, quality and model versions.
- They do not establish how common rising or falling costs were across applications.
Evidence (2 items)
Context only: Cursor,
fairly constant
Why it matters here: Cursor confirms substantial cost differences between requests while reporting broadly stable costs for most users. This supports caution about token consumption but does not establish either an industry-wide rise or fall.
- Publisher
- Cursor
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Context only: OpenAI,
26% less expensive
Why it matters here: OpenAI reports lower median-query costs for a newer model. This challenges an unrestricted denial of cost reductions, but its typical query assumptions do not measure spending across representative applications.
- Where
- Pricing
- Publisher
- OpenAI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
What would settle it
Representative longitudinal application records for August 25, 2023–August 25, 2025 showing inference costs per comparable unit of completed work, accounting for input, output, reasoning and cached tokens, model changes, quality, task volume and difficulty.
Where it was said (1)
Said once in the sources this register covers.
: Where's Your Ed At, How To Argue With An AI Booster
The cost of inference has gone up. Statements otherwise are purely false, and are the opinion of somebody who does not know what he's talking about.
- Where
- paragraph 236
- Published
- 2025-08-25
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment wyea-how-to-argue-with-an-ai-booster-15df5f-a2: Current. Outcome: Unresolved. 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-how-to-argue-with-an-ai-booster-15df5f-a2.