OpenAI Next-Gen Model 'Sol' Leaked with 6x Speed Improvement Claims

Leaked benchmark reports concerning OpenAI’s next-generation model, codenamed Sol, indicate inference speeds roughly six times faster than existing baselines. Beyond the raw performance claims, the report details an evolving workflow inside OpenAI's research divisions, where staff members routinely supervise multiple autonomous AI assistants during daily experiments. Although unverified officially, the leak underscores a sustained push toward drastic latency reduction and agent-driven internal tooling.