OpenAI Launches GPT-6.1 Sol with Near-Astra Performance at One-Fifth of the Price

If you’ve ever looked at the pricing for a top-tier AI model and thought “that’s not for the likes of us,” OpenAI’s latest announcement might be worth a second look. On 29 September 2026, the company posted details of GPT-6.1 Sol — a new model sitting below its flagship GPT-6 Astra raw capability, but priced at one-fifth of Astra’s standard input and output token costs. For developers, businesses and organisations that run AI at scale, that gap matters enormously.

The announcement landed through OpenAI’s official account and product materials, framing the model as an upgrade to the existing GPT-6 Sol rather than a brand-new product line. But the pricing difference is striking enough that it’s already generating attention in developer circles.

What Is GPT-6.1 Sol, Exactly?

Put simply, it’s OpenAI’s answer to a question a lot of businesses have been asking: can we get close to the best performance without paying the top-tier price?

GPT-6.1 Sol is available through the OpenAI API under the model identifier `gpt-6.1-sol`. OpenAI describes it as suited to “repeated, long-running work” — think software engineering, document processing, business workflows, computer use and scientific reasoning. It’s not the company’s most powerful model. That remains GPT-6 Astra. But OpenAI says GPT-6.1 Sol delivers near-Astra performance for the kinds of complex, sustained tasks where many organisations actually spend most of their AI budget.

The technical specs are sizeable. OpenAI’s developer documentation lists a context window of 1.05 million tokens — meaning the model can hold an enormous amount of text in its working memory at once — and a maximum output of 128,000 tokens per response. Its knowledge cut-off is 30 April 2026.

How Much Does It Cost?

Standard API pricing, as published by OpenAI, is $2 per million input tokens (around £1.55 at current rates), $0.10 per million cached input tokens (around £0.08), and $10 per million output tokens (around £7.75). For longer contexts, those figures rise to $4 per million input tokens, $0.20 per million cached input tokens, and $15 per million output tokens.

The cached-input price deserves particular attention. OpenAI says it sits 95% below the model’s standard input price and 50% below the cached-input price of the previous GPT-6 Sol. For applications that repeatedly reuse the same large prompt or document — a legal firm processing similar contracts, say, or a developer running the same codebase through repeated checks — that reduction could meaningfully cut running costs.

Bear in mind that these prices are published in US dollars. Sterling costs will shift with exchange rates, and UK users would also need to factor in VAT and any applicable commercial terms. OpenAI has not separately published UK sterling pricing.

Who Can Use It Right Now?

At launch, GPT-6.1 Sol is available to Plus, Pro, Business, Enterprise and Edu users through ChatGPT Work and Codex. OpenAI confirmed the model was not yet available in the standard ChatGPT Chat service at the time of the announcement.

OpenAI describes GPT-6.1 Sol as “the most cost-efficient model for its performance available today.” That’s a company claim, and it’s worth being clear: independent verification of that assertion across all competing models and workloads has not been established. The AI market is busy — Google, Anthropic, Meta and others all have models in this performance and price range — and benchmark results cited by OpenAI are company-reported figures covering selected tests in software engineering, business workflows, computer use and scientific reasoning.

What the Developers Are Saying

No named OpenAI executive quote accompanied the launch materials at the time of writing. But the developer community’s interest has centred on the cached-input pricing in particular. Applications that hold large, reused contexts — code repositories, lengthy documents, ongoing business workflows — stand to benefit most from that 95% reduction on cached tokens.

Sam Altman, OpenAI’s chief executive, has previously said the company’s goal is to make powerful AI accessible to as many people and organisations as possible. GPT-6.1 Sol’s pricing structure appears to reflect that direction, though whether it achieves it in practice will depend heavily on the specific workloads organisations bring to it.

Benchmarks and Independent Assessment

OpenAI’s published comparisons cover software engineering, business workflows, computer use and scientific reasoning. The company has not, at the time of writing, provided a full head-to-head comparison with every competing model at equivalent price points. Developers and organisations considering the model should treat the benchmark figures as a starting point for their own evaluation rather than a definitive verdict.

What This Means for Kent Residents

For Kent businesses, universities and public-sector organisations — including those in professional services, technology, education and health — GPT-6.1 Sol’s lower pricing could make API-based AI tools more financially accessible for software development, document processing and administrative automation, provided they meet OpenAI’s access requirements and can account for dollar-denominated costs, exchange-rate effects and VAT. Any organisation in Kent handling personal, confidential or NHS-related data would need to work through data protection, information governance and cybersecurity requirements carefully before deploying the model in a live service. As with all AI tools at this stage, the question isn’t just whether you can afford to use it — it’s whether you’ve done the groundwork to use it responsibly.