OpenAI’s new GPT-5.6 Sol model independently post-trained the smaller Luna model using a prompt the company described as “fairly underspecified.”
How the automation works
The goal for AI labs is to use artificial intelligence to accelerate their own development cycles. OpenAI claims GPT-5.6 Sol achieves this better than any previous system.
After Luna underwent its initial pre-training phase, Sol took over to optimise the model for specific skills and behaviours without human intervention. A researcher provided Sol with instructions through the Codex platform. The task required the model to identify correct training configurations, select appropriate GPUs, launch the training script, and verify that everything ran correctly.
“Previously this is something that a team of senior researchers may have worked on at OpenAI, and now it really feels like the automated researcher is pretty close,” said Kathy Shi, an OpenAI researcher, during the presentation.
Sol beats GPT-5.5 by 16 points on self-improvement benchmark
OpenAI built an internal evaluation suite based on real-world research tasks to measure these abilities directly. The suite covers debugging research systems, optimising kernels and training recipes, running machine learning experiments, and improving other models.
GPT-5.6 Sol scores 16.2 points higher than GPT-5.5 on the aggregated Recursive Self-Improvement (RSI) index. Sol sits at the top of the benchmark’s model hierarchy, followed by the Terra and Luna variants, then GPT-5.5 and GPT-5.4.
Recursive Self-Improvement refers to an AI system’s ability to make itself better, where each round of gains makes the system even more capable of improving itself. That creates a feedback loop. The term has long been central to AI safety research because a system that can recursively improve itself could, in theory, trigger a rapid explosion in capability.
OpenAI rival Anthropic stressed in early June that full recursive self-improvement has not been achieved yet but “could come sooner than most institutions are prepared for.” Full RSI means an AI system that designs its own successor without human help. According to Anthropic, Claude can now handle incremental work between major paradigm shifts, and humans are responsible for only a single-digit percentage of directional decisions.
Token output per researcher more than doubles with GPT-5.6
OpenAI says its researchers use GPT-5.6 Sol across the entire development cycle, from debugging and optimising training systems to running experiments and reading results. Even during internal testing, average daily token output per active researcher more than doubled the previous peak set by GPT-5.5. Pull requests and experiments per researcher went up too, letting teams turn ideas into results faster.
The company’s own adoption numbers from the past six months paint a predictably rosy picture. The share of compute allocated to internal coding inference grew 100x, while agent-based token usage jumped roughly 22x. OpenAI acknowledges these metrics do not directly measure research progress but says they show how fast AI-assisted work is scaling.
What it means
For the people building these tools, the change is practical rather than theoretical. Tasks that previously required a senior researcher to spend days configuring hardware and debugging pipelines can now happen automatically. This frees up human time for higher-level strategy and reduces the friction of setting up experiments. The shift moves the burden of execution from the researcher to the model, allowing teams to iterate faster without needing to micromanage the technical setup.
