AI researchers at Stanford and the University of Washington trained an AI 'reasoning' model named s1 for under $50 using cloud compute credits. The model, which performs similarly to OpenAI’s o1 and DeepSeek’s R1, is available on GitHub. It was developed using distillation from Google’s Gemini 2.0 Flash Thinking Experimental model and demonstrates strong performance on benchmarks.
This speculative article explores the idea that GPT-5 might already exist internally at OpenAI but is being withheld from public release due to cost and performance considerations. It draws parallels with Anthropic's handling of a similar situation with Claude Opus 3.5, suggesting that both companies might be using larger models internally to improve smaller models without incurring high public-facing costs. The author examines the potential motivations behind such decisions, including cost control, performance expectations, and strategic partnerships.
