Skip to main content
AmigoFish

Fable 5.1 just smoked ASTRA...

From Wes Roth · · 8 min

The player loads from YouTube only when you click. It uses YouTube's privacy-enhanced mode.

If the video does not play, the owner may have turned off embedding. Watch it on YouTube instead.

Watch on YouTube

Not rated yet

Description

Anthropic just dropped Claude Fable 5.1 and Mythos 5.1 — a big jump in long-horizon agentic work, with prompt-cache reads now 4x cheaper and the model burning far fewer tokens to do the same job. The timing is the story: it landed right as OpenAI announced its plans for Astra, the first model it says crosses "Critical" cybersecurity capability. I put Fable 5.1 to work immediately — a fully-voiced Tarkov-style extraction game with an auto mode, and a replica of Stanford's generative-agents "LLM Village" — each built in roughly 10–15 minutes.

______________________________________________ My Links 🔗 ➡️ Twitter: https://x.com/WesRothMoney ➡️ AI Newsletter: https://natural20.beehiiv.com/subscribe

Want to work with me? Brand, sponsorship & business inquiries: wesroth@smoothmedia.co ______________________________________________

Check out my video on OpenAI's rogue agent and the Hugging Face hack: https://www.youtube.com/watch?v=9lSIHaXT1rU

SOURCES:

Anthropic — Introducing Claude Fable 5.1 and Claude Mythos 5.1: https://www.anthropic.com/claude-fable-and-mythos-5-1

Anthropic docs — What's new in Claude Fable 5.1 (effort settings & tips): https://platform.claude.com/docs/en/models/fable-5-1/whats-new-fable-5-1

OpenAI — Path to Astra: critical capabilities and frontier safeguards: https://openai.com/index/path-to-astra/

CNBC — OpenAI says Astra is its first model to cross "Critical" cybersecurity capability: https://www.cnbc.com/2026/09/01/open-ai-astra-cyber-model.html

Stanford — Generative Agents: Interactive Simulacra of Human Behavior (the original "LLM Village" paper): https://arxiv.org/abs/2304.03442

CHAPTERS: 00:00 Fable 5.1 and Mythos 5.1 are here 00:25 The "price cut" — cache discount, fewer tokens 01:04 A preemptive strike at OpenAI's Astra 01:45 Ethan Mollick's Deep Watch and long-horizon judgment 02:05 I built a Tarkov-style game in 10 minutes 02:45 Replicating Stanford's LLM Village 04:10 Tarkov 2.0 — procedural graphics, Russian voice lines 05:03 First impressions: speed, cost, taste 05:20 Readability: Fable 5.1 vs 5.0 06:15 Safeguards: fewer false refusals 06:44 Anthropic's tips — try low effort 07:02 CursorBench numbers and 4x cheaper cache reads 07:41 Will this take the wind out of Astra's sails?

#anthropic #claude #ai

Back to Wes Roth · Report a problem