An interesting perspective on the Anthropic pitch that you should be writing loops instead of prompting Claude.
A comparison of Fable vs GLM vs GPT shows Fable outperforms other LLMs at long running tasks but is slower and more expensive for quick tasks.
