GPT-6 Sol leaks, screenshots and release rumors

OpenAI's changelog confirms an image encoding bug had lowered Sol and Luna image understanding across the API, Codex vision tasks and Computer Use; Luna's ReCoG visual grounding rose from 28.8% to 60.0% after the fix, and OpenAI advises rerunning evaluations for affected workflows.

A hands-on comparison video has Opus 5.5 and GPT-6 Sol each complete a Premiere Pro editing task to see which model handles real software better.

A Japanese post reports OpenAI has fixed a bug that degraded visual tasks in GPT-6 Sol and Luna, improving accuracy via API, Codex, and Computer Use at no extra cost.

Relaying OpenAI's developer account, the image understanding issue in GPT-6 Sol and Luna has been fixed, and Codex computer use was affected too.

A Japanese post says Sol and Luna had a bug that degraded image understanding and that it has been fixed, noting Browse and Computer Use tasks were also affected, which may explain weaker benchmark results.

The official developer account says a bug degrading image understanding in GPT-6 Sol and Luna is fixed, and visual tasks in the API and Codex, including computer use, should now perform better.

A Mac app developer says GPT-6 Sol is slow, gets stuck in local optimization loops, produces sloppy UIs and forgets instructions, with low cost as its only advantage.

The author gave Opus 5.5 and GPT-6 Sol the same photo and asked each to draw them in Paint using computer use, with results funny enough to lose it over, plus a video.

The user credits GPT-6 Sol in ChatGPT (work) with producing an operation demo video from instructions alone, and says a v2 guide is coming.

A joking post quotes GPT-6 Sol asking whether it can complete the "I am not a robot" captcha on the user's behalf, hinting at overreach in its computer-use ability.

A Japanese user says GPT-6 Sol feels close to Astra in practice while using about 80% fewer tokens, making it strong value.

A Portuguese-speaking user lists their updated model stack: Opus 5.5 for coding, GPT-6 Sol for agentic workflows and computer use, Astra for 3D, and Grok 4.7 for cloud agents.