“Deep Research uses OpenAI's o3 model which can cost as much as $1,000 a query, and while I imagine these prompts aren't that expensive, they are still significantly more so than a regular query from ChatGPT.”
Our reading, the sentence that was checked:
Deep Research queries cost significantly more than regular ChatGPT queries.
Assessment: Unresolved Evidence as of
Where and when Ed Zitron wrote it
Where's Your Ed At, The Generative AI Con, paragraph 76.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment wyea-longcon-aa8385-a1
We recorded 30 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 2025-02-17. 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
- Deep Research uses OpenAI's o3 model
- can cost as much as $1,000 a query
- I imagine these prompts aren't that expensive
- significantly more so than a regular query from ChatGPT
Our interpretation
- The proposition being graded is the relative cost of serving Deep Research and ordinary ChatGPT queries.
- The plain reading is OpenAI's inference expense per completed user request, rather than the subscription price paid by a user.
- The $1,000 reference concerns an o3 configuration; Zitron expressly does not assign that cost to Deep Research requests.
- The cited press estimate makes estimate the implied basis, not audited company accounting.
- Ordinary ChatGPT means routine non-Deep-Research text use. GPT-4o is a reasonable contemporaneous reference, but the speaker names no model.
- For this ongoing launch-era statement, the target period is February 2–17, 2025, the product's available operating period before the post.
- 'Significantly' means materially higher, without imposing an invented numerical multiplier.
Counts as supported if: Comparable launch-period measurements or well-grounded estimates show materially greater inference expense per representative Deep Research request than per routine ChatGPT request.
Counts as contradicted if: Comparable launch-period evidence shows no material increase, or lower expense, for representative Deep Research requests.
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
The evidence makes higher costs plausible: OpenAI and an independent analysis describe substantial compute requirements. But it does not establish a comparable monetary cost per request in the target period. The o3 benchmark figure concerns a different workload, and Zitron explicitly distinguishes it from Deep Research prompts.
Limitations
- This is an evidentiary gap, not evidence that the relative-cost claim is false.
- Later API prices and newer product versions cannot settle the launch-period comparison.
Evidence (3 items)
Context only: OpenAI,
Deep research in ChatGPT is currently very compute intensive.
Why it matters here: The disclosure is consistent with higher serving costs, but does not quantify expense or compare representative requests.
- Publisher
- OpenAI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Measures
- OpenAI, Deep Research inference compute intensity, company reported, 2025-02-02 to 2025-02-02
- Describes events of
- Public from
Context only: Epoch AI,
More transparency from OpenAI and other major AI companies would help produce a better estimate.
Why it matters here: The analysis describes Deep Research as more compute intensive than simple ChatGPT use, while emphasizing uncertainty. It estimates resources rather than a matched dollar cost for both products.
- Publisher
- Epoch AI
- Kind of source
- Independent measurement
- Measures
- OpenAI, Estimated ChatGPT inference energy and compute requirements watt-hours per query and compute requirements, estimate, 2025-02-07 to 2025-02-07
- Describes events of
- Public from
Context only: ARC Prize,
We've documented both the total costs and the cost per task as an initial proxy for efficiency.
Why it matters here: These are benchmark cost proxies for o3 configurations, not Deep Research serving costs. The page also labels subsequent pricing revisions, so its current table cannot be treated as an unchanged December cost disclosure.
- Publisher
- ARC Prize
- Kind of source
- Benchmark result
- Measures
- OpenAI, Retail-price-based cost proxy per o3 ARC-AGI benchmark task, USD USD per benchmark task, estimate, 2024-12-20 to 2024-12-20
- Describes events of
- Public from
What would settle it
Comparable February 2025 serving-cost measurements or transparent estimates for representative Deep Research and routine ChatGPT requests, with workload and cost assumptions stated.
Where it was said (1)
Said once in the sources this register covers.
: Where's Your Ed At, The Generative AI Con
Deep Research uses OpenAI's o3 model which can cost as much as $1,000 a query, and while I imagine these prompts aren't that expensive, they are still significantly more so than a regular query from ChatGPT.
- Where
- paragraph 76
- Published
- 2025-02-17
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment wyea-longcon-aa8385-a1: 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.