Simon Willison compares the current state of large language models to the newly opened Jurassic Park, noting that ignoring them is as futile as a geneticist dismissing dinosaur DNA. The site lists recent developments including a price war between Claude Opus 5.5 and GPT-6 variants, the introduction of System One Decision Models by Jev, and the generation of running routes using GPT-6 Astra and ChatGPT Work. These entries appear on the 22nd, 21st, and 12th of September 2026 respectively, suggesting a period of intense activity and competition within the sector.
The comparison highlights a specific tension between scientific skepticism and market reality. While some experts might view the technology as merely frog DNA or a marketing stunt involving human casualties, the sheer volume of new model releases and pricing strategies indicates otherwise. The industry continues to move forward regardless of whether the underlying mechanisms are truly novel or simply repackaged existing capabilities.
- Claude Opus 5.5 faces direct competition from GPT-6 Sol and GPT-6 Luna.
- Jev launches System One, described as a new shape of LLM called Decision Models.
- Tools like GPT-6 Astra and ChatGPT Work now generate specific running routes.
