Codex Opens GPT-5.6 Sol's 1M-Token Context to ChatGPT Accounts
OpenAI's Tibo Sottiaux said on August 17, 2026 that GPT-5.6 Sol's 1M-token context option in Codex now works through ChatGPT accounts, not only API keys. OpenAI's docs list a 1,050,000-token API window for the model and a model_context_window setting, but no ChatGPT-account rule. GPT-6.1 Sol is now Codex's recommended model.
OpenAI's Tibo Sottiaux said on August 17, 2026 that GPT-5.6 Sol's 1M-token context option in Codex now works for usage through ChatGPT accounts, not only API keys (Tibo Sottiaux on X). OpenAI's documentation backs the pieces around that claim but not the account rule itself: the GPT-5.6 Sol model page lists a 1,050,000-token context window, and the Codex configuration reference documents the model_context_window setting.
GPT-5.6 Sol has a successor in Codex. OpenAI's models page now recommends GPT-6.1 Sol for complex coding and agentic workflows (ChatGPT docs: Models).
Key Takeaways
- Sottiaux says the 1M-token option works with ChatGPT accounts. He says it previously worked only with API keys. OpenAI's Codex docs do not document this account rule, so treat it as an OpenAI developer's statement, not a documented guarantee.
- The model supports it in the API. OpenAI's GPT-5.6 Sol page lists a 1,050,000-token context window and 922,000 maximum input tokens.
- The default context length has not been replaced. Sottiaux says it was tuned for performance and cost.
- More context costs more allowance. OpenAI's Codex pricing docs say sessions that hold more context use significantly more per message.
- GPT-6.1 Sol is the current recommendation. OpenAI's models page recommends it for complex coding in Codex.
What Changed
The access boundary is the main change Sottiaux described. He says GPT-5.6 Sol 1M in Codex previously worked only for API keys, then that the switch had been flipped for usage through ChatGPT accounts as well. (Tibo Sottiaux on X)
The model side is documented. OpenAI's API model page lists GPT-5.6 Sol with a 1,050,000-token context window, 922,000 maximum input tokens, and 128,000 maximum output tokens, and prices API prompts over 272K input tokens at 2x input and 1.5x output for the full request. (OpenAI GPT-5.6 Sol model page)
What OpenAI's Codex docs do not say is which account types, plans, or clients get the larger window. The configuration reference describes model_context_window only as the "Context window tokens available to the active model". (Codex configuration reference)
Why the Default Still Matters
Sottiaux was explicit that a larger context window is not automatically better for every task. He says the current context length is the default because it was tuned for performance and cost, while the 1M-token option is available for users who need it. (Tibo Sottiaux on X; Tibo Sottiaux on X)
OpenAI's Codex pricing docs make the cost side concrete: larger projects, long-running tasks, or extended sessions "that require the agent to hold more context will use significantly more per message." (ChatGPT docs: Pricing) Treat the larger window as a workload-specific setting rather than a universal upgrade.
What to Verify Before Enabling It
Sottiaux's August 16 post walks through enabling a 1M-token window in Codex. The documented setting is model_context_window in config.toml; model_auto_compact_token_limit sets the token threshold that triggers automatic history compaction. (Codex configuration reference) Before using it on a real project, check:
- Whether your Codex client and account actually report the larger window after a new session starts.
- How the larger window changes latency and usage for your repository.
- Whether the task actually needs more retained context than the default.
- Whether the workflow has a clear way to recover if a long context becomes unwieldy.
Successor: GPT-6.1 Sol in Codex
OpenAI released GPT-6.1 Sol on September 29, 2026. As of October 27, 2026, OpenAI's models page says GPT-6.1 Sol offers near-Astra performance at a lower cost than Astra, that Luna suits focused, repeatable tasks, and that Plus, Pro, Business, Enterprise, and Edu plans should choose GPT-6.1 Sol as the GPT-5.5 replacement. (ChatGPT docs: Models) The pricing page still lists GPT-5.6 Sol, Terra, and Luna credit rates as unchanged, with GPT-5.6 Sol's promotional pricing available at least through November 21, 2026. (ChatGPT docs: Pricing) Its API model page lists the same 1,050,000-token context window as GPT-5.6 Sol. (OpenAI GPT-6.1 Sol model page)
Bottom Line
OpenAI's Tibo Sottiaux says GPT-5.6 Sol's 1M-token context option works through ChatGPT accounts. OpenAI documents the model's 1,050,000-token window and the model_context_window setting, but not the account rule, so confirm the effective window in your own session. The default stays tuned for performance and cost, and for new work OpenAI now recommends GPT-6.1 Sol in Codex.
Corrections and Updates
- October 9, 2026: This page rested only on posts on X and did not mention a successor. It now cites OpenAI's GPT-5.6 Sol model page, Codex configuration reference, and pricing docs, says OpenAI's docs do not document the ChatGPT-account rule, and points to GPT-6.1 Sol.
Sources
- Tibo Sottiaux on X, GPT-5.6 Sol 1M in Codex for ChatGPT accounts (August 17, 2026): https://x.com/thsottiaux/status/2089143488696705077
- Tibo Sottiaux on X, instructions for enabling the 1M-token Codex context window (August 16, 2026): https://x.com/thsottiaux/status/2089082893804896524
- OpenAI API docs, GPT-5.6 Sol model page: https://developers.openai.com/api/docs/models/gpt-5.6-sol
- OpenAI API docs, GPT-6.1 Sol model page: https://developers.openai.com/api/docs/models/gpt-6.1-sol
- ChatGPT docs, Codex configuration reference: https://learn.chatgpt.com/docs/config-file/config-reference
- ChatGPT docs, Pricing (usage limits): https://learn.chatgpt.com/docs/pricing
- ChatGPT docs, Models: https://learn.chatgpt.com/docs/models
Read next
Related coverage- OpenAI
OpenAI Previews GPT-5.6: Sol, Terra, and Luna in Limited Preview
OpenAI previewed the GPT-5.6 family in June 2026 with trusted partners: Sol, a frontier flagship it called a step function better than GPT-5.5; Terra, competitive with GPT-5.5 at 2x lower cost; and Luna, its most cost-efficient model. All three became generally available on July 9, 2026, and GPT-6 has since superseded them.
- Codex
Codex Can Import Workflows From Other AI Agents
OpenAI now lets the ChatGPT desktop app and Codex CLI import supported setup and recent work from other agents. The desktop app can import from Claude Code, Claude Cowork, and Cursor, while Codex CLI supports Claude Code and Cursor, with automatic updates and import history available in the desktop app.
- ChatGPT
ChatGPT Expands GPT-5.6 Luna Access and Adds a Reasoning Slider
In August 2026 OpenAI made GPT-5.6 Sol the single Chat experience for Plus and Pro, with a reasoning slider on web, mobile, and desktop, and moved Free and Go to GPT-5.6 Luna with unlimited text chats and a Think button. Since October 7, GPT-6 Sol and Luna are rolling out in their place.
- OpenAI
OpenAI Expands Daybreak With GPT-5.6-Cyber
OpenAI expanded its Daybreak cybersecurity program with Blue and Red access tiers and introduced GPT-5.6-Cyber, a purpose-trained model for authorized vulnerability research, exploit validation, and security testing. Access is limited to approved defenders and organizations.
Frequently Asked Questions
What changed for GPT-5.6 Sol in Codex?
OpenAI's Tibo Sottiaux said on X on August 17, 2026 that GPT-5.6 Sol's 1M-token context option in Codex, which had worked only with API keys, now works through ChatGPT accounts too. OpenAI's Codex documentation does not state this account rule.
Is the 1M-token Codex context window the default?
No. Sottiaux said Codex's context limit is tuned for performance and cost and stays the default. OpenAI's Codex configuration reference documents model_context_window, the setting for the context window tokens available to the active model, and its pricing docs say holding more context uses more allowance.
How large is GPT-5.6 Sol's context window?
OpenAI's GPT-5.6 Sol model page lists a 1,050,000-token context window, 922,000 maximum input tokens, and 128,000 maximum output tokens in the API. In the API, prompts over 272K input tokens are priced at 2x input and 1.5x output for the whole request.
Which model should I use in Codex now?
OpenAI's models page recommends GPT-6.1 Sol for complex coding and agentic work, and Luna for focused, repeatable tasks. GPT-6.1 Sol is in Codex for Plus, Pro, Business, Enterprise, and Edu, and its API page also lists a 1,050,000-token context window.
Get an email when Codex changes its plans, prices or limits
We check every change against Codex's own pages before we send. You also get the weekly AI Catchup; one click unsubscribes.