---
format: "aidr-story-markdown/v1"
id: "abab33e242d348d9dc713d6833ab1a93a054f4f1501225aba6d5efa1a8941078"
canonical_url: "https://aidr.today/abab33e2?lang=en"
title: "METR: 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)"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-11-05T19:01:21.000Z"
category: "Research"
topics: ["anthropic","llm","safety"]
source_urls: ["https://www.techmeme.com/261105/p46","https://huggingnews.com/ai/metr-audit-finds-anthropic-understated-plausibility-of-ai-randd-accelera-8f53d391","https://marketbrief.now/ai/metr-audit-finds-anthropic-understated-plausibility-of-ai-randd-accelera-8f53d391","https://huggingnews.com/ai/anthropics-ai-model-development-speed-nearly-doubles-as-research-capabil-c443fff8","https://marketbrief.now/ai/anthropics-ai-model-development-speed-nearly-doubles-as-research-capabil-c443fff8","https://huggingnews.com/ai/metr-says-anthropic-ai-progress-may-have-hit-2x-scaling-safety-threshold-b191f768","https://marketbrief.now/ai/metr-says-anthropic-ai-progress-may-have-hit-2x-scaling-safety-threshold-b191f768","https://news.ycombinator.com/item?id=50383902"]
summary: "METR reports that Anthropic's models are improving about twice as fast as they did between March 2024 and October 2025. Internal AI use is driving approximately 80% of this acceleration."
---

# METR: 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 \(Tiffany Li/The Information\)

> [Open the canonical story](<https://aidr.today/abab33e2?lang=en>)

**Published:** 2026-11-05T19:01:21.000Z
**Category:** Research
**Topics:** anthropic, llm, safety

## Summary

METR reports that Anthropic's models are improving about twice as fast as they did between March 2024 and October 2025\. Internal AI use is driving approximately 80% of this acceleration\.

## Sources

- [Story source](<https://www.techmeme.com/261105/p46>)
- [Story source](<https://huggingnews.com/ai/metr-audit-finds-anthropic-understated-plausibility-of-ai-randd-accelera-8f53d391>)
- [Story source](<https://marketbrief.now/ai/metr-audit-finds-anthropic-understated-plausibility-of-ai-randd-accelera-8f53d391>)
- [Story source](<https://huggingnews.com/ai/anthropics-ai-model-development-speed-nearly-doubles-as-research-capabil-c443fff8>)
- [Story source](<https://marketbrief.now/ai/anthropics-ai-model-development-speed-nearly-doubles-as-research-capabil-c443fff8>)
- [Story source](<https://huggingnews.com/ai/metr-says-anthropic-ai-progress-may-have-hit-2x-scaling-safety-threshold-b191f768>)
- [Story source](<https://marketbrief.now/ai/metr-says-anthropic-ai-progress-may-have-hit-2x-scaling-safety-threshold-b191f768>)
- [Discussion](<https://news.ycombinator.com/item?id=50383902>)