Science & Tech · AI · 5 hrs ago
Researchers say Anthropic is developing AI about twice as fast as last year
Researchers at nonprofit METR say Anthropic is developing its AI systems about twice as quickly as a year ago.
The estimate concerns models’ ability to perform tasks with clearly defined goals, not every kind of work.
METR reviewed company data and interviewed staff during an assessment lasting about 11 weeks.
The researchers believe AI tools themselves are helping speed up development; Anthropic has said its Claude chatbot helps build newer versions.
The finding contrasts with CEO Dario Amodei’s call for the industry to slow down so safety measures can keep pace with improving AI.
The results matter because questions remain about AI risks and human control, while companies have largely supplied information about their own technology and safety.
METR says its estimate has caveats, and Anthropic’s redactions limit how others can check some details; the company is also preparing to go public later this month.
METR, an independent nonprofit research group, says Anthropic is improving its AI systems for well-specified tasks around twice as quickly as it was a year ago.
The group said AI assistance was likely a major reason for the faster development.
Its assessment drew on Anthropic’s internal data and staff interviews over about 11 weeks.
METR cautioned that its findings could be overstated and that Anthropic’s redactions limited the evidence it could publish.
The findings come as CEO Dario Amodei has called for the AI industry to slow its work so safety measures can catch up.
- Who
- Anthropic and Model Evaluation and Threat Research (METR), an independent nonprofit research group.
- What
- METR estimated that Anthropic is improving its models for well-specified tasks around twice as quickly as a year earlier.
- When
- The report was released Thursday, November 5, 2026. METR reviewed data from mid-August through late October.
- Where
- METR is based in Berkeley, California; Anthropic is based in San Francisco.
- Why
- METR assessed whether Anthropic’s internal capability growth had accelerated and how much AI contributed to that acceleration.
This story does not have two clearly opposing sides.
Dario Amodei published an essay calling on the AI industry to slow the pace of improvement.
METR analyzed internal data provided by Anthropic and interviewed employees.
METR released its analysis estimating that Anthropic’s rate of improvement had doubled.
- Estimated pace
- Around twice as fast as a year earlier for well-specified tasks
- Research group
- Model Evaluation and Threat Research (METR)
- Data review period
- About 11 weeks, from mid-August through late October
- METR finding
- Anthropic made as much progress on benchmarks in the past year as from October 2023 to October 2025
- Unreleased models
- METR said Anthropic had finished training three models in the previous six months
5 sources
3 takeaways from the big AI report on Anthropic everyone is talking about
Anthropic’s AI development has doubled in speed despite slowdown calls, research group says
Anthropic is rapidly accelerating AI development amid CEO's call to slowdown, group says
Anthropic is rapidly accelerating AI development amid slowdown calls, outside research group says
Anthropic is speeding up its AI development, outside researchers say
What people are saying
slow the pace at which we improve the capabilities of AI models.
AI is speeding up our research, but it’s hard to say exactly how much
While we were able to freely share our highest-level conclusions about risk, Anthropic redacted a majority of the evidence we found most informative for these conclusions
Now only bottlenecked on ideas & compute, no longer bottlenecked on code