· Sep 1, 2026
Epoch AI points readers to its analysis of an ECI trend break
Markdown · JSON · Original source
’s page, published September 1, 2026, points readers to a report about detecting a trend break in the ECI and some other measures of AI capability with . It directs readers to the report for details of the methodology and lists downloadable model ECI scores, trend lines and bootstrap samples. The supplied page text does not state the trend’s rate or explain what the trend break means for AI capabilities.
Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.
Original source
· Article
Source details
Original source date: Sep 1, 2026 (2026-09-01T00:00:00.000Z).
Date basis: Date supplied by the publisher’s feed.
Added to Escape Meridian: 2026-11-06T05:08:24.840Z. First observed: 2026-10-18T02:55:25.940Z. Retrieved: 2026-10-23T21:23:37.835Z.
Referenced by
Recorded citations and public file attachments. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
Text as retrieved on October 23, 2026; originally published September 1, 2026. The supplied page text does not include the report’s methodology, the graph, or its data, so it does not establish the trend’s rate or implications.
Open question
What methodology did use to detect the trend break?
Source-stated
Epoch AI says its report gives details of the methodology used to detect a trend break in the ECI and some other measures of AI capability with reasoning models.
See our report “Have AI capabilities accelerated?” for details on the methodology we used to detect that the ECI (and some other measures of AI capability) had a trend break with reasoning models.
Source-stated
Epoch AI lists downloadable model ECI scores, trend lines and bootstrap samples.
Model ECI scores CSV, Updated Sep. 1, 2026 Reasoning trend line and 90% prediction interval
Source and contributor context
Epoch AI · Publisher
The publisher is identified here. No individual author has been inferred.
Source media
Media retrieval status: unchecked.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Reasoning model
A language model trained to work through a problem in written steps before it answers.
Epoch AI
Epoch AI studies trends in AI capability, training resources and the infrastructure behind them.
Editorial provenance
Version 1. Publication ID: f054d6a2-a5ec-55c2-feae-8f1f9b9dd1ac.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-06T05:04:09.426Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model.
Observation: da745bd3cfed1f4ba23bd990c8c60d1fc4c34750cceacfd4b805362301bb63d4. Source text hash: 7ad0b2c59da68b46c9efd4ef4cb6eb5166503a8d4a5ce0968f89f2e39df2231a.
Revision and source-check history
Version 1 · 2026-11-06T05:08:24.840Z
Epoch AI points readers to its analysis of an ECI trend break
’s page, published September 1, 2026, points readers to a report about detecting a trend break in the ECI and some other measures of AI capability with . It directs readers to the report for details of the methodology and lists downloadable model ECI scores, trend lines and bootstrap samples. The supplied page text does not state the trend’s rate or explain what the trend break means for AI capabilities.
Original source
· Article
Source details
Original source date: Sep 1, 2026 (2026-09-01T00:00:00.000Z).
Date basis: Date supplied by the publisher’s feed.
Added to Escape Meridian: 2026-11-06T05:08:24.840Z. First observed: 2026-10-18T02:55:25.940Z. Retrieved: 2026-10-23T21:23:37.835Z.
Limits
Text as retrieved on October 23, 2026; originally published September 1, 2026. The supplied page text does not include the report’s methodology, the graph, or its data, so it does not establish the trend’s rate or implications.
Open question
What methodology did use to detect the trend break?
Source-stated
Epoch AI says its report gives details of the methodology used to detect a trend break in the ECI and some other measures of AI capability with reasoning models.
See our report “Have AI capabilities accelerated?” for details on the methodology we used to detect that the ECI (and some other measures of AI capability) had a trend break with reasoning models.
Source-stated
Epoch AI lists downloadable model ECI scores, trend lines and bootstrap samples.
Model ECI scores CSV, Updated Sep. 1, 2026 Reasoning trend line and 90% prediction interval
Source and contributor context
Epoch AI · Publisher
The publisher is identified here. No individual author has been inferred.
Source media
Media retrieval status: unchecked.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Reasoning model
A language model trained to work through a problem in written steps before it answers.
Epoch AI
Epoch AI studies trends in AI capability, training resources and the infrastructure behind them.
Editorial provenance
Version 1. Publication ID: f054d6a2-a5ec-55c2-feae-8f1f9b9dd1ac.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-06T05:04:09.426Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model.
Observation: da745bd3cfed1f4ba23bd990c8c60d1fc4c34750cceacfd4b805362301bb63d4. Source text hash: 7ad0b2c59da68b46c9efd4ef4cb6eb5166503a8d4a5ce0968f89f2e39df2231a.
uncertain · 2026-11-06T05:04:09.426Z · anthropic/claude-opus-5.5
Publication: f054d6a2-a5ec-55c2-feae-8f1f9b9dd1ac
attribution — pass: The reading attributes the page and its pointer to the methodology report to Epoch AI, the publisher. The text is one data-insight page, not a mere listing, though the retrieved text appears to lack the page's main body.
See our report “Have AI capabilities accelerated?” for details on the methodology we used
summary — uncertain: The central term ECI is never expanded or explained in the reading, and it has no entry in the supplied definitions. A general reader cannot tell what is being measured. 'Trend break' is also left unexplained. Only 'reasoning model' is defined. Smallest repair: add a term-library entry for ECI, plus a plain-language gloss of 'trend break' as a change in the rate of improvement. The reading correctly notes that the page text gives no rate or implications.
the ECI (and some other measures of AI capability) had a trend break with reasoning models
date — uncertain: The displayed date of September 1, 2026 comes from a backfilled feed date. The only dates in the page text are data files marked 'Updated Sep. 1, 2026', which may describe a later revision rather than the page's first publication. The related items dated Dec. 23, 2025 and Apr. 16, 2026 are earlier, and the reading does not rely on them. There is no hindsight framing and no forecast presented as an outcome. The first-publication date remains unconfirmed.
Model ECI scores CSV, Updated Sep. 1, 2026
support — pass: Both claims match the text verbatim: the pointer to the methodology report, and the downloadable ECI scores, trend lines and bootstrap samples. The limitations accurately note what is missing, and the question follows from the evidence without adding unsupported implications.
Reasoning trend line and 90% prediction interval CSV, Updated Sep. 1, 2026
media — pass: No media was selected, and no verified media candidates were supplied, so nothing is misidentified.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.
Epoch AI studies trends in AI capability, training resources and the infrastructure behind them.
Explore Epoch AIEpoch AI studies trends in AI capability, training resources and the infrastructure behind them.
Explore Epoch AIA language model trained to work through a problem in written steps before it answers.
Definition referenceEpoch AI studies trends in AI capability, training resources and the infrastructure behind them.
Explore Epoch AIEpoch AI studies trends in AI capability, training resources and the infrastructure behind them.
Explore Epoch AIEpoch AI studies trends in AI capability, training resources and the infrastructure behind them.
Explore Epoch AIA language model trained to work through a problem in written steps before it answers.
Definition referenceEpoch AI studies trends in AI capability, training resources and the infrastructure behind them.
Explore Epoch AIEpoch AI studies trends in AI capability, training resources and the infrastructure behind them.
Explore Epoch AI