Wolfram Ravenwolf
@WolframRvnwlf
AI Evangelist @ @CoreWeave/@wandb, @thursdai_pod co-host. Opinions my own.
Evaluates models for breakfast, builds agents at night, preaches AI all day long. 😎
参加 September 2023
1.3K フォロー中 4.7K ファン
Spent $11K evaluating Claude Fable 5 on @WolfBenchAI.
It had #1# potential, but outright refusals dragged its final score down.
Surprising result: Fable does not even surpass Opus 4.6.
もっと見る
