OpenAI fixed a bug in image encoding in the GPT-6 Sol and GPT-6 Luna models
OpenAI has fixed a bug in image encoding in the GPT-6 Sol and GPT-6 Luna models that was degrading visual input understanding in both the API and the Codex product. The company recommends re-verifying affected workflows.
OpenAI fixed a bug in image encoding that was degrading visual input understanding in the GPT-6 Sol and GPT-6 Luna models. The fix was deployed on 25 September 2026 and concerns the use of these models both via the API and in the Codex product.
According to the company, the fix improves results on tasks involving visual content, including computer use scenarios. The company recommends that users who use image inputs in their workflows rerun their evaluations and repeat tasks that may have been affected by the bug.
See the source article for details.
Why it matters
Users who relied on the visual understanding of these models (image recognition, computer use in the Codex product) may have been receiving systematically worse results without knowing about the bug. After the fix, it is appropriate to re-check output quality to determine whether earlier issues were actually related to this bug or were a different kind of failure.
Two audiences, two different impacts
What this means
For individuals
Developers who used image inputs with the GPT-6 Sol or GPT-6 Luna models may have encountered worse results due to a bug in image encoding; after the fix, it is advisable to retest your tasks involving visual content.
For a business
Companies running pipelines with image inputs via the GPT-6 Sol or GPT-6 Luna models (API, Codex, computer use) may have received worse results on visual tasks due to a bug; it is now advisable to re-verify output quality and, if needed, recompute the affected workflows.
DevelopmentCheck the original
Event sources
clearly official source · 1 publisher, 0 independent. We count feeds from the same owner only once.