Claude 3 is out and Anthropic claim it is the most intelligent language model on the planet. The paper was released 90 minutes ago, and I’ve read it in full and the release notes. I’ve tested the model and compared it to Gemini 1.5 and GPT-4 in image analysis, business use cases, long context, logic, mathematics, JSON outputting, risqué content, creative writing, official benchmarks and more.

In short, I think the model will be popular … but why so, and what does that mean for AGI?

AI Insiders:   / aiexplained  

Claude 3 Opus: https://claude.ai/chats
Paper, w/ Opus, Sonnet and Haiku: https://www-cdn.anthropic.com/de8ba9b...
Release Notes: https://www.anthropic.com/news/claude...
Pricing, Opus, Sonnet and Haiku: https://www.anthropic.com/api#pricing
Amodei Interview: https://www.dwarkeshpatel.com/p/dario...
NYT Anthropic: https://www.nytimes.com/2023/07/11/te...
LLM Leaderboard: https://huggingface.co/spaces/lmsys/c...
Gemini 1.5: https://storage.googleapis.com/deepmi...
GPQA: https://arxiv.org/pdf/2311.12022.pdf
GPT-4 Turbo Benchmark, Kinda: https://arxiv.org/html/2401.02985v1

AI Insiders:   / aiexplained  

Non-Hype, Free Newsletter: https://signaltonoise.beehiiv.com/