OpenAI has raised the default context length of GPT-6.1 Sol to 500,000 tokens when the model is used through Codex and ChatGPT Work, Codex lead Thibault Sottiaux wrote on X late on Sunday, 11 October. The previous default was 272,000 tokens; he described the change as “an ~ 80% increase”.
Sottiaux said users can still override the default, and that the model’s maximum context length is higher. “We will go higher in the future,” he wrote.
The post was short. “We have increased default context length for GPT-6.1 Sol to 500k when used through Codex & Work, up from 272k or an ~ 80% increase,” Sottiaux wrote.
He gave two reasons. The larger default, he wrote, “helps reduce occurrence of compactions and with tasks that require keeping a lot of context and nuance over periods of time.”
He also said the number was chosen with care: “It was carefully tested and is the new sweet spot for most of our users.” By our arithmetic the change adds 228,000 tokens to the default.
GPT-6.1 Sol’s default context in Codex and ChatGPT Work rose from 272,000 to 500,000 tokens, per Codex lead Thibault Sottiaux’s X post of 11 October 2026, shown against the 1,050,000-token context window listed on OpenAI’s model page. Zero-based, in thousands of tokens.
Business Pill · THE CONTEXT WINDOW
A short explainer of what a context window is: how much text an AI model can keep in view at once. It teaches the general idea only and says nothing about any company in this story.
The key insight: As we read it, the change moves where a long Codex session starts, not how far it can go: the ceiling was already above 1,000,000 tokens, and Sottiaux’s post raises the starting point to 500,000. The old 272,000 default matched the point where OpenAI’s API price list begins charging long-context rates.
Default, Override and Maximum
The 500,000 figure is a default, not a ceiling. “You can still override it if you want to be higher our lower, with maximum context length of GPT-6.1 Sol of 1M,” Sottiaux wrote.
OpenAI’s model page for GPT-6.1 Sol lists a 1,050,000 context window and 128,000 max output tokens. On that page’s figure, the new default is a little under half of the model’s full window, by our arithmetic.
The chart below sets the old default, the new default and the model page’s context window side by side.
Where 272,000 Also Appears
The old default matches a threshold on OpenAI’s API price list. The model page says: “Prompts with more than 272K input tokens are priced at 2x input and cache rates and 1.5x output for the full request.”
The page lists standard API prices of $2.00 per million input tokens, $0.10 per million cached input tokens and $10.00 per million output tokens. By our arithmetic, a request above 272,000 input tokens is therefore priced at $4.00 per million input tokens and $15.00 per million output tokens across the whole request.
Sottiaux’s post is about the default in Codex and ChatGPT Work. It does not mention API pricing, and it does not say whether the API’s long-context rates changed.
Standard GPT-6.1 Sol API prices from OpenAI’s model page, read 12 October 2026, and the price of a request whose prompt exceeds 272K input tokens: our arithmetic from the page’s note (2x input and cache rates, 1.5x output, for the full request).
What Compaction Means Here
Sottiaux tied the change to “compactions”. OpenAI’s developer guide describes the mechanism: “To support long-running interactions, you can use compaction to reduce context size while preserving state needed for subsequent turns.” The same guide says compaction “helps you balance quality, cost, and latency as conversations grow.”
Sottiaux’s post makes three points. The first is the number: the default in Codex and ChatGPT Work rose from 272,000 to 500,000 tokens, which he called an ~80% increase.
The second is the reason he gave: fewer compactions, and better handling of tasks that need a lot of context over time. OpenAI’s compaction guide describes compaction as reducing context size while preserving the state needed for later turns.
The third is that the default is a choice users can change. He wrote that users can set it higher or lower, up to a maximum he put at 1M, and that OpenAI will go higher in the future.
Thibault Sottiaux, who leads OpenAI Codex, on X (11 October 2026)
“It was carefully tested and is the new sweet spot for most of our users.”
Three Implications
CODEX USERS Sottiaux wrote that the larger default helps reduce occurrence of compactions, so in his account long sessions can run further before earlier context is condensed.
API BUILDERS The model page says prompts above 272K input tokens are priced at 2x input and cache rates and 1.5x output for the whole request; the post does not say those rates changed.
TEAMS SETTING LIMITS He wrote that users can still override the default in either direction, so the 500,000 figure is a starting point rather than a cap.
The analysis puts it this way: “Orchestration, memory, tool use, and workflow execution have become the new sources of competitive advantage.”
As we read it, a default context length is one of those harness settings: OpenAI changed what a Codex session keeps in view without changing the model’s maximum.
What Is Not Established
The post does not say what testing OpenAI ran, which tasks it measured, or how often compaction happened before and after the change. It gives no date for a further increase beyond “in the future”. It does not say whether the change reaches ChatGPT plans other than Work, or whether usage limits or credits shift with the larger default. His “1M” maximum and the model page’s 1,050,000 context window are both reported as each source states them.
Business Engineer Framework
The Agentic Harness War
A Business Engineer analysis of how the layer around the model, from memory to tools to workflow, became the place AI products compete.
OpenAI now starts GPT-6.1 Sol in Codex and ChatGPT Work with a 500,000-token context, up from 272,000, according to the Codex lead’s post. He said the aim is fewer compactions on long tasks, that users can still set the length themselves, and that the default will go higher.
The old default sat at the same 272,000-token mark where OpenAI’s API starts charging long-context rates. The post does not say whether that threshold played any part in the change.
A note on sourcing. On 12 October 2026 we read in full Thibault Sottiaux’s X post of 11 October, OpenAI’s GPT-6.1 Sol model page and OpenAI’s compaction guide, all linked below. The defaults, his reasons and the further increase come from his post; the context window, output limit and prices come from the model page. The 228,000-token gap, the share of the full window and the long-prompt prices are our arithmetic.
Some reports of the change carried further details about subscriptionpricing and input limits that we could not match to these sources, so we left them out. Nothing here is investment advice.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur, HackerNoon and Search Engine People. Find him on LinkedIn and Substack. See how we source.
Scroll to Top
Discover more from FourWeekMBA
Subscribe now to keep reading and get access to the full archive.
