Skip to content
Unfold Byte UB markUBUnfoldByte
Model watch

METR report: internal AI use drove 80% of Anthropic’s model acceleration

A nonprofit study suggests Anthropic’s model improvements doubled in speed, with internal AI tools accounting for most of the gain.

Unfold Byte Newsroom

Depth
80%Internal AI usage drove approximately 80% of this acceleration, according to the published report The findings come from a report by the AI research nonprofit METR · reported
1 outlet · newest source 4h ago

Key points

  • Internal AI usage drove approximately 80% of this acceleration, according to the published report
  • The findings come from a report by the AI research nonprofit METR.
METR report: internal AI use drove 80% of Anthropic’s model acceleration

The claim

Tiffany Li of The Information reported these findings, noting that internal AI adoption may have significantly doubled the pace of development at Anthropic.

Evidence: reportedMETR: Anthropic's models are improving about twice as fast as they did between March 2024 and October 2025, with internal AI use driving ~80% of acceleration (Tiffany Li/The Information)

aifleet-techmeme

The response

The source material does not contain any direct response, rebuttal or comment from Anthropic regarding the METR report. There are no quoted statements or attributed counterpoints from company executives or researchers in the provided text. The report stands as an independent assessment by the nonprofit without recorded pushback from the subject company.

The bottom line for teams

The data suggests that integrating AI into internal workflows can substantially accelerate model development cycles. However, the report does not detail specific tools or methodologies used by Anthropic. Teams should view this as a potential benchmark for efficiency gains rather than a prescriptive guide, given the lack of operational detail in the source material.

What the paper shows

Why it matters

  • Model evaluators can use this data to benchmark expected improvement rates against internal AI integration levels.
  • The 80% attribution highlights the potential ROI of AI-driven development processes for research teams.
  • Understanding acceleration drivers helps evaluators distinguish between architectural changes and process efficiencies.
  • The lack of Anthropic’s response means evaluators must treat these figures as external estimates rather than confirmed metrics.

Topics: models, Anthropic, METR, AI development, research

Updated

The strongest doubt

Counterpoint from the sources

The source material provides no opposing viewpoint or doubt from Anthropic or other researchers.

Ask the article

Local keyword search of this article; no generated answers.

Did this article give you the evidence you needed?

No vote saved on this device.
Your choice stays on this device. No public tally.

Questions readers ask

What drove most of Anthropic’s model acceleration?

Internal AI use drove approximately 80% of the acceleration, according to the METR report.

Did Anthropic comment on the report?

The source material does not include any response or comment from Anthropic.

Connected stories

This storyAnthropic’s updated filing reveals compute costs rose to 58% of Q3 revenueAnthropic’s updated filing reveaWill unified robots.txt signals standardise AI training opt-outs?Will unified robots.txt signals Anthropic discloses $895.4bn compute commitments in amended SEC filingAnthropic discloses $895.4bn comCan AI solve ninety-year-old math problems and render schooling obsolete?Can AI solve ninety-year-old mat

Reading receipt

  • Internal AI usage drove approximately 80% of this acceleration, according to the published report
  • The findings come from a report by the AI research nonprofit METR.

Still open: How exactly did internal AI tools contribute to the 80% acceleration?

Still open: Will other vendors see similar gains from internal AI adoption?

Time spent: 0:00 · Depth: Full

Sources

  1. Techmeme

Related stories