Skip to content
PROVENANCE BRIEF
Research

Academic or research source. Check the methodology, sample size, and whether it's been replicated.

Even frontier LLMs from GPT-5 onward lose up to 33% accuracy when you chat too long

Even with newer models like GPT-5.2 and Claude 4.6, AI chatbots still give worse answers the longer a conversation goes on.

2-Minute Brief
  • Affects widely-used AI models.
  • Even with newer models like GPT-5.2 and Claude 4.6, AI chatbots still give worse answers the longer a conversation goes on.
  • The article Even frontier LLMs from GPT-5 onward lose up to 33% accuracy when you chat too long appeared first on The Decoder .
  • Open receipts to verify and go deeper.
8-Minute Deep Dive

Context

Even with newer models like GPT-5.2 and Claude 4.6, AI chatbots still give worse answers the longer a conversation goes on. The article Even frontier LLMs from GPT-5 onward lose up to 33% accuracy when you chat too long appeared first on The Decoder .

For builders

Even with newer models like GPT-5.2 and Claude 4.6, AI chatbots still give worse answers the longer a conversation goes on.

Verify

Prefer primary announcements, papers, repos, and changelogs over reposts.

Read Original

Even frontier LLMs from GPT-5 onward lose up to 33% accuracy when you chat too long

TLDR

Even with newer models like GPT-5.2 and Claude 4.6, AI chatbots still give worse answers the longer a conversation goes on.

2-Minute Brief
  • Affects widely-used AI models.
  • Even with newer models like GPT-5.2 and Claude 4.6, AI chatbots still give worse answers the longer a conversation goes on.
  • The article Even frontier LLMs from GPT-5 onward lose up to 33% accuracy when you chat too long appeared first on The Decoder .
  • Open receipts to verify and go deeper.
8-Minute Deep Dive

Context

Even with newer models like GPT-5.2 and Claude 4.6, AI chatbots still give worse answers the longer a conversation goes on. The article Even frontier LLMs from GPT-5 onward lose up to 33% accuracy when you chat too long appeared first on The Decoder .

For builders

Even with newer models like GPT-5.2 and Claude 4.6, AI chatbots still give worse answers the longer a conversation goes on.

Verify

Prefer primary announcements, papers, repos, and changelogs over reposts.

Open
O open S save B back M mode