Anthropic's Next AI Infers Vague Intent for October Release

AI1 day ago

The San Francisco AI lab is developing a model that can identify a broad class of software bugs after being tasked with fixing only one, The Information reports. The unreleased system is designed to deduce a user's underlying objectives when instructions are imprecise and to solve problems more creatively than existing commercial models. It is expected to launch in October, coinciding with the release of Google's Gemini 4 Carbon and a new Terra model from OpenAI.

The reporting follows leaks of a 'Fable 5.5' model in Microsoft Azure and LiteLLM's GitHub data. Traces of a model named 'claude-onigiri-eap-2' also appeared in Artificial Analysis's code, featuring four distinct reasoning levels from low to 'xhigh.' While Anthropic has not officially announced a date, several sources indicate corporate partners have already tested two separate checkpoints of the Fable 5.5 model.

Mapped to your portfolio

Possible scenarios

40%Strong October Launch+0.4%
30%Guarded Safety Rollout+0.2%
20%Release Slips-0.3%

Each figure is the modeled move of the S&P 500 if the outcome occurs. For information only. Not investment advice.

Looking for more context or analysis?

Tilt chat provides reliable, tailored insights, alerts, and automations.

Anthropic's Next AI Infers Vague Intent for October Release

In this story

Anthropic Google OpenAI

Portfolio

SPY
State Street SPDR S&P 500 ETF Trust
100%

Portfolio

SPY
State Street SPDR S&P 500 ETF Trust
100%

Mapped to your portfolio

Possible scenarios

40%Strong October Launch+0.4%
30%Guarded Safety Rollout+0.2%
20%Release Slips-0.3%

Each figure is the modeled move of the S&P 500 if the outcome occurs. For information only. Not investment advice.