Boil Alert

Influencer claims Llama 3 ends the model race.

A stealth startup claims a private demo proved general intelligence.

Reports claim o1 solves international math olympiad problems reliably.

Social media claims the open Llama 3 70B matches GPT-4 quality at no cost.

Critics claim SD3 produces unsafe content at scale.

Vendor-adjacent blog claims DALL-E 3 eliminates the need for designers.

Blog post claims Midjourney v6 makes stock photography obsolete.

OpenAI announces o1, a model that thinks before answering.

Anthropic claims Claude 3 Opus writing is indistinguishable from expert humans.

Vendor claims SOTA open-weight image generation.

Google claims Gemini can ingest a full film and answer questions about it.

Google DeepMind announces a 1M-token context window.

Anthropic reports Claude 3 Opus surpasses GPT-4 on MMLU and graduate-level QA.

OpenAI reports GPT-4 scores in the top 10% of bar exam takers.

Vendor claims GPT-4 reaches human-level performance on coding benchmarks.