GPT-5.1: What UK businesses need to know

Whats changed inside ChatGPT

GPT-5.1 replaces GPT-5 in the ChatGPT interface for paid users.

The key improvements:

• Instant is smarter. It keeps the speed of the old Instant model but now decides when to think more deeply. This helps with tricky questions, data checking and quick analysis.
• Thinking is clearer. It adjusts thinking time based on task difficulty. It gives more structured answers and fewer unclear technical statements.
• Tone is better. Responses are warmer and more natural, which helps in client-facing tasks, sales copy and team communication.
• No change to context limits. Instant stays at 32k tokens. Thinking stays at 196k tokens.

Alongside the model upgrades, ChatGPT now offers stronger personalisation controls. Users can set the tone of the assistant with options including Professional, Friendly, Candid, Quirky, Efficient, Nerdy and Cynical. You can also fine-tune how concise, warm or scannable responses should be. These settings apply instantly across all chats, even ongoing ones. ChatGPT will also suggest adjustments when it notices you asking for a specific style. This makes it easier for teams to match ChatGPT’s voice to their brand, their workflows or their communication standards.

These improvements make ChatGPT more useful for analysts, consultants, marketers, sales teams and founders who rely on the UI for day-to-day work.

Important limitation for teams

Testing in our workspaces confirms that the new model breaks the @mention feature for Assistants and Custom GPTs.
You cannot currently type @{AssistantName} to bring another assistant into the thread.
Teams that rely on multi-assistant workflows will need to switch to manual copy and paste or use the API until OpenAI fixes it.

What changes for API users

The API continues to use the GPT-5 family, with the same pricing and context limits.

Key points for technical teams:

• No new pricing. Costs stay at £1.25 per million input tokens and £10 per million output tokens for GPT-5.
• Same context window. 400k total tokens, with around 272k usable input in practice.
• Same structure. You pick from GPT-5, GPT-5-mini or GPT-5-nano depending on cost and performance needs.
• GPT-5.1 behaviour is a ChatGPT upgrade, not a new API model. Engineering teams should continue using the existing GPT-5 endpoints.

This means UK companies already building on the API do not need to rewrite anything. The main gains are in the ChatGPT experience, not the underlying platform.

What UK businesses should do next

• Update internal guidance so staff know Instant is now stronger and can handle more analytical work.
• Use Thinking for long documents, technical analysis and complex reasoning.
• Review workflows that relied on @mentions, as these are currently not supported in GPT-5.1.
• Keep API usage unchanged unless you want to optimise for cost by switching to mini or nano models.

GPT-5.1 is an incremental upgrade, not a full shift in capability. The benefit is steady, reliable improvements in how teams work with ChatGPT every day.