Simon Willison on Nostr: nprofile1q…dxsqp the very best vision models - Clause 3.5 Sonnet, Google Gemini 1.5 ...
nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpqrs3jj0wtfz0jvphlldx68xsy2qdw63204gulkxlhn0yrpe05l5cqsdxsqp (nprofile…xsqp) the very best vision models - Clause 3.5 Sonnet, Google Gemini 1.5 Pro, GPT-4o - can produce massively more detailed and world-knowledge backed descriptions then the tint 4.2GB file that runs on my laptop
I still wouldn't trust them for an ornithological level of accurate detail though - maybe a model will get there in the next 12 months but I try not to lean too much into guessing future abilities
I still wouldn't trust them for an ornithological level of accurate detail though - maybe a model will get there in the next 12 months but I try not to lean too much into guessing future abilities