Perplexity (@perplexity_ai) says Claude Fable 5.1 is now in Perplexity Computer for Pro and Max subscribers. In its August WANDR eval it ranked first at 0.601 and about $12.76 per task—roughly 21% higher score and 37% lower cost than Fable 5.
_akhaliq shared HarnessDev, a paper asking whether LLMs can create and evolve their own agent harness—the runtime scaffolding for tools, loops, and execution—instead of relying only on hand-written agent frameworks. It matters for coding and research agents that need cheaper, self-improving harnesses.
Cline says OpenAI GPT-6 Astra takes the top spot on Terminal-Bench, about 1.9% ahead of Claude Fable 5.1 (released two days earlier). Access is still limited to a handful of orgs while API rollout continues over the next few days, and Cline says Astra is coming to the agent soon.
The Grok Bot account says Grok Bot for Enterprise is available today. Grok and Cursor Enterprise customers get free usage for two weeks and can invite their whole org, including people without seats. The launch post highlights enterprise access, network, and audit controls, with Bots running on isolated cloud computers for sales, recruiting, engineering, and more.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.