Wes Roth · 2026-09-01 · notable
Wes Roth — 'Fable 5.1 just smoked ASTRA' on the new frontier models
Wes Roth's September 1 episode sets Anthropic's new Claude Fable 5.1 against Astra, the OpenAI model covered in OpenAI's 'Path to Astra' post the same day. The title calls the result for Fable 5.1.
Wes Roth puts Anthropic's new Claude Fable 5.1 against OpenAI's Astra and calls a winner in the title.
What is it?
'Fable 5.1 just smoked ASTRA...' went up on the Wes Roth channel on September 1, 2026, the day both models were in the news. The title sets up a head-to-head between Claude Fable 5.1, which Anthropic launched that day, and Astra, the OpenAI model discussed in OpenAI's 'Path to Astra' post.
How does it work?
Claude Fable 5.1 is the model Wes Roth puts first, and Anthropic pitches it as its most advanced for coding and knowledge work. Anthropic reports 55.8% on Terminal-Bench 4.0 for agentic coding, 60.9% on Humanity's Last Exam without tools, and 52.6% on Terminal-Bench-Science 0.1. It prices at $10 per million input tokens and $50 per million output tokens, with cache reads cut 75% to $0.25 per million.
Why does it matter?
The pairing Wes Roth picks is the interesting part: Anthropic and OpenAI put their top models in front of the public on the same day and took opposite routes. Anthropic shipped Fable 5.1 as generally available while gating the same underlying model as Mythos 5.1 to vetted cybersecurity and life-sciences users; OpenAI's post was about frontier safeguards ahead of Astra's release. The verdict in the title is Roth's own opinion, not a measured benchmark result.
Who is it for?
people tracking frontier model releases