OpenAI Increases GPT-6.1 Sol Default Context Length to 500k
OpenAI has increased the default context length of the GPT-6.1 Sol model in Codex and Work environments from 272k to 500k tokens. Additionally, OpenAI confirmed that no extra long-context fees will be charged for this expanded window. This roughly 80% increase significantly reduces the frequency of context compression and helps developers handle complex tasks requiring large amounts of historical data. Users benefit from improved long-context handling without incurring higher subscription usage costs. While 500k tokens is now the default sweet spot, users can still manually adjust the context length up to a maximum upper limit of 1M tokens. The update applies specifically to Codex and Work environments.
## BACKGROUND
GPT-6.1 Sol is a large language model released by OpenAI. Context length in LLMs refers to the maximum number of tokens—words or sub-words—that a model can process and retain in its immediate attention window during a single interaction.