OpenAI: GPT-6 Sol

GPT-6 Sol is a text model from OpenAI, part of the GPT-6 series. Within that series it sits below the flagship GPT-6 Astra and above the fast GPT-6 Luna tier, positioned as the cost-efficient high-end option. It was released by the provider on 2026-09-22. If you are choosing between tiers, Sol is the middle option: less capable than Astra, more capable than Luna.

On the input side, the model accepts text, images and files, and it produces text as output. The context window is 1,050,000 tokens, which is large enough to hold very long documents, whole codebases or extended conversation histories in a single request. The maximum output is 128,000 tokens, so responses can themselves be long, not just the inputs. Because images and files are accepted alongside text, you can mix formats in one prompt rather than converting everything to plain text first.

Those capabilities point to a few practical patterns. Long-context work is the obvious one: reviewing lengthy contracts, specifications or log files without chunking them into separate calls. Mixed-modality prompts suit tasks where a screenshot, diagram or scanned page needs to be read together with surrounding text, such as extracting structured data from a document that includes figures. The large output ceiling helps for generating long reports, migrations or multi-file code changes in one response. Tool calling and structured outputs make it usable as the reasoning step inside an agent loop or a pipeline that needs machine-readable results rather than prose.

You call it through api14's OpenAI-compatible API using the model id "openai/gpt-6-sol". The supported request parameters are include_reasoning, max_completion_tokens, max_tokens, reasoning, reasoning_effort, response_format, seed, structured_outputs, tool_choice, tools and verbosity. In practice that means you can cap output length, control how much reasoning the model does and how it is surfaced, request JSON or schema-conforming responses, set a seed for reproducibility, define tools and choose how they are selected, and adjust response verbosity. If you already have code written against the OpenAI chat completions shape, switching to this model is mostly a matter of changing the model id and adding whichever of these parameters you need.

Keep the output limit in mind: 128,000 tokens is generous but finite, so tasks that need to emit more than that in one pass have to be split. The model outputs text only, so anything requiring generated images or audio needs a different model. And since Sol is positioned below GPT-6 Astra, workloads where the highest available capability matters more than efficiency are the ones where Astra may fit better.