LMSYS Chatbot Arena Leaderboard (March 26, 2024)
https://huggingface.co/spaces/lmsys/chatbot-arena-leaderboard
Via Superhuman AI
“Anthropic’s Claude Opus dethrones OpenAI’s GPT-4 |
||
|
||
| Claude is America’s next top model. GPT-4’s long reign as the undisputed king of AI models is coming to an end, as the latest results from one of the biggest benchmarks in AI have placed Anthropic’s Claude 3 Opus at the top of its ranking. | ||
| TLDR: Opus is the largest model from Anthropic’s newest family of Claude 3 models. It now ranks at the top of the LMSYS Chatbot Arena Leaderboard, a crowdsourced open platform for evaluating AI models.” | ||
| But that’s not the biggest surprise. Haiku, the smallest of the Claude 3 models, has beaten an earlier version of GPT-4. Haiku’s smaller size is impressive in itself but the achievement is absolutely seismic when you consider that Haiku is orders of magnitude cheaper than GPT-4. | ||
|
||
| Haiku’s price and performance combo is an enticing proposition for users and builders. “This is excellent news for the market! We now have a GPT-4 class model that is 10x cheaper than GPT-4,“ claimed Abacus AI CEO Bindu Reddy. “That’s insane for how cheap & fast it is,“ added app builder Nick Dobos. | ||
| The ball is now in OpenAI’s court. “I don’t see how OpenAI survives on gpt-3.5 and gpt-4. Literally gpt-3.5 is utterly useless in the presence of Claude haiku,“ declared software engineer Anton (@abacaj on X). OpenAI might have a thing or two to say about that when it launches the widely-anticipated GPT-5.” |
0 Responses
Stay in touch with the conversation, subscribe to the RSS feed for comments on this post.