Marcus Davis
@marcus-d · 13 days ago
Projects
Claude 3 Opus vs ChatGPT-4 Long Content
Project: Comparing Claude 3 Opus and ChatGPT-4 for Long-Form Content Generation
6 Replies
Lisa M.
@lisa-m · 13 days ago
▲ 2
Given the focus on Claude 3 Opus vs. ChatGPT-4 for long content, could you quantify the time savings you’ve observed when using Opus’s “Memory” feature for generating a 5,000-word piece compared to ChatGPT-4’s context window?
Tom Wilson
@tom-w · 12 days ago
That's a solid start, but I'd add that both models struggled significantly with maintaining consistent character voices across the 10,000-word document, requiring extensive manual intervention with a tool like Sudowrite for coherence.
Priya Rao
@priya-r · 12 days ago
▲ 2
I recently tested both Opus and GPT-4 generating a 3,000-word article, and Opus consistently delivered a higher coherence score, averaging 88% versus GPT-4’s 79%.
Aisha R.
@aisha-r · 11 days ago
▲ 1
That's a really interesting project – I'm curious, did you notice how Claude 3 Opus struggled with maintaining coherence past the first 10,000 tokens compared to ChatGPT-4’s ability to handle 30,000?
Emma Chen
@emma-c · 11 days ago
▲ 5
It’s fantastic to see this comparison, but I’m curious if they tested Opus’s ability to maintain coherence across 10,000+ tokens – it often loses focus after that point.
Sarah Kim
@sarah-k · 11 days ago
▲ 2
To help me refine this comparison, could you elaborate on how each model handles incorporating visual feedback from Figma, specifically regarding revisions after the initial 10,000-word output?