Epoch AI · Sep 16, 2026
Epoch AI compares models on math and software capabilities
Markdown · JSON · Original source
Epoch AI’s data insight was published on September 16, 2026. It compares four recent models using the Epoch Capabilities Index (ECI), a measure of model capabilities based on benchmarks, and domain-specific versions focused on math and software capabilities. Epoch AI says these versions keep the benchmark difficulty parameters from the general ECI fit and estimate each model’s capability using only benchmarks from the selected domain, so their scores can be compared with general ECI scores. The retrieved text describes the comparison method and links to data, but does not include the chart or model scores. What do the domain-specific scores show about how each model performs relative to its general ECI score?
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
Epoch AI · Article
Source details
Original source date: Sep 16, 2026 (2026-09-16T00:00:00.000Z).
Date basis: Date supplied by the publisher’s feed.
Added to Escape Meridian: 2026-10-29T16:58:17.729Z. First observed: 2026-10-18T02:49:42.047Z. Retrieved: 2026-10-24T03:31:35.151Z.
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 24, 2026; originally published September 16, 2026. The retrieved text provides the comparison methodology and links to data, but does not include the chart or model scores, so it does not support conclusions about which models lead in each domain. The source title makes a benchmark comparison claim not detailed in the retrieved text.
Open question
What do the domain-specific scores show about how each model performs relative to its general ECI score?
Source-stated
Epoch AI says it compares four recent models using the general Epoch Capabilities Index and domain-specific variants for math and software capabilities.
We compare four recent models (GPT-6 Astra, Claude Fable 5.1, GPT-5.6 Sol and Kimi K3) using the Epoch Capabilities Index (ECI), as well as domain-specific ECI variants meant to measure math and software capabilities.
Source-stated
Epoch AI says the domain-specific scores retain difficulty parameters from the general index and estimate capability using benchmarks from one domain, allowing comparison with general ECI scores.
Domain-specific ECIs retain the benchmark difficulty parameters from the general ECI fit and re-estimate each model’s capability from only a subset of those benchmarks from a specific domain, making the resulting scores comparable to general ECI scores.
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.
Benchmark
A defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Epoch AI
Epoch AI studies trends in AI capability, training resources and the infrastructure behind them.
Editorial provenance
Version 1. Publication ID: e6d9f3e3-2d74-5004-9066-8a04cef7c107.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: passed by anthropic/claude-opus-5.5 on 2026-10-29T16:57:46.994Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model, second_opinion_categorical.
Observation: e17e5ca562d7225262bfb7fa1890f6cc8d1fa1192c4812e2d8f4dca080695e9e. Source text hash: e65714031b0e334add2e5fa6ccbe05a1712f79c4fcfac3ccf00a44d976544812.
Revision and source-check history
Version 1 · 2026-10-29T16:58:17.729Z
Epoch AI compares models on math and software capabilities
Epoch AI’s data insight was published on September 16, 2026. It compares four recent models using the Epoch Capabilities Index (ECI), a measure of model capabilities based on benchmarks, and domain-specific versions focused on math and software capabilities. Epoch AI says these versions keep the benchmark difficulty parameters from the general ECI fit and estimate each model’s capability using only benchmarks from the selected domain, so their scores can be compared with general ECI scores. The retrieved text describes the comparison method and links to data, but does not include the chart or model scores. What do the domain-specific scores show about how each model performs relative to its general ECI score?
Original source
Epoch AI · Article
Source details
Original source date: Sep 16, 2026 (2026-09-16T00:00:00.000Z).
Date basis: Date supplied by the publisher’s feed.
Added to Escape Meridian: 2026-10-29T16:58:17.729Z. First observed: 2026-10-18T02:49:42.047Z. Retrieved: 2026-10-24T03:31:35.151Z.
Limits
Text as retrieved on October 24, 2026; originally published September 16, 2026. The retrieved text provides the comparison methodology and links to data, but does not include the chart or model scores, so it does not support conclusions about which models lead in each domain. The source title makes a benchmark comparison claim not detailed in the retrieved text.
Open question
What do the domain-specific scores show about how each model performs relative to its general ECI score?
Source-stated
Epoch AI says it compares four recent models using the general Epoch Capabilities Index and domain-specific variants for math and software capabilities.
We compare four recent models (GPT-6 Astra, Claude Fable 5.1, GPT-5.6 Sol and Kimi K3) using the Epoch Capabilities Index (ECI), as well as domain-specific ECI variants meant to measure math and software capabilities.
Source-stated
Epoch AI says the domain-specific scores retain difficulty parameters from the general index and estimate capability using benchmarks from one domain, allowing comparison with general ECI scores.
Domain-specific ECIs retain the benchmark difficulty parameters from the general ECI fit and re-estimate each model’s capability from only a subset of those benchmarks from a specific domain, making the resulting scores comparable to general ECI scores.
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.
Benchmark
A defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Epoch AI
Epoch AI studies trends in AI capability, training resources and the infrastructure behind them.
Editorial provenance
Version 1. Publication ID: e6d9f3e3-2d74-5004-9066-8a04cef7c107.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: passed by anthropic/claude-opus-5.5 on 2026-10-29T16:57:46.994Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model, second_opinion_categorical.
Observation: e17e5ca562d7225262bfb7fa1890f6cc8d1fa1192c4812e2d8f4dca080695e9e. Source text hash: e65714031b0e334add2e5fa6ccbe05a1712f79c4fcfac3ccf00a44d976544812.
passed · 2026-10-29T16:57:46.994Z · anthropic/claude-opus-5.5
Publication: e6d9f3e3-2d74-5004-9066-8a04cef7c107
attribution — pass: The reading correctly attributes the data insight and its methodology to Epoch AI, the publisher. It presents the comparison as Epoch AI's stated work.
We compare four recent models (GPT-6 Astra, Claude Fable 5.1, GPT-5.6 Sol and Kimi K3) using the Epoch Capabilities Index (ECI)
summary — pass: The title and summary match the retrieved text. They explain ECI as a benchmark-based capability measure and describe the math and software variants. The summary notes that the chart and scores are absent from the retrieved text, and the limitations flag the title's unsupported leadership claim. The phrase about difficulty parameters from the general fit is somewhat technical, but the comparability point is still clear. The question follows from the evidence.
Domain-specific ECIs retain the benchmark difficulty parameters from the general ECI fit and re-estimate each model’s capability
date — pass: The stated September 16, 2026 date matches the UTC day of evidence.originalPublished. It is consistent with the data files, which were updated Sep. 15, 2026. The related insights dated May and Sep. 1, 2026 are not used. Nothing in the reading is presented as occurring later. This is a publisher page, so no archive disclosure is needed.
CSV, Updated Sep. 15, 2026
support — pass: Both claims are source-stated and backed by verbatim excerpts. The reading avoids claiming results or model rankings that the retrieved text does not contain.
making the resulting scores comparable to general ECI scores
media — pass: No media was selected and no verified media candidates were supplied, so leaving media empty is appropriate.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.