GPT-6.1 Sol Review: Features, Benchmarks & Pricing

GPT-6.1 Sol is OpenAI's upgrade to GPT-6 Sol, released at DevDay on September 29, 2026 — just one week after the original. OpenAI's pitch is simple: near-Astra performance on agentic coding, computer use, and professional work, at one-fifth of GPT-6 Astra's standard token price and exactly the same price as GPT-6 Sol.

This GPT-6.1 Sol review covers what changed from GPT-6 Sol, what the benchmark numbers say, what it costs, and where its limits sit. It also covers access: GPT-6.1 Sol is not yet integrated into Chat Smith, but GPT-6 Sol is already live there, with limited use on the free plan and higher usage limits on Chat Smith Pro.

GPT-6.1 Sol is the middle model of OpenAI's GPT-6 family, sitting between GPT-6 Astra and GPT-6 Luna. It replaces GPT-6 Sol as OpenAI's recommended mid-tier model in the API, with the same 1,050,000-token context window, up to 128,000 tokens of output, and support for text and image input.

OpenAI says the model uses the same kinds of data and training as GPT-6 Astra. Reasoning effort runs across five levels from low to max, with medium as the default. Unlike the original GPT-6 Sol, it has no option to switch reasoning off entirely.

There is one catch for everyday users: GPT-6.1 Sol is not available in ChatGPT's regular Chat tab yet. For now it lives in ChatGPT Work, Codex, and the API, while GPT-6 Sol still powers GPT-6 in Chat for paid ChatGPT users.

Four areas show the biggest gains over GPT-6 Sol.

On DeepSWE v1.1, which tests software engineering in real codebases, GPT-6.1 Sol matches GPT-6 Astra at roughly one-fifth of the cost per task and scores 6.4 points higher than GPT-6 Sol. For coding agents and long development sessions, that considerably narrows the case for paying Astra prices.

On GDP.pdf, which asks professional questions about dense PDFs full of tables, charts, diagrams, and fine print from finance, healthcare, legal, and other fields, GPT-6.1 Sol beats Claude Opus 5.5 with fallbacks at less than half the cost per task. It also scores 4.8 points higher than GPT-6 Sol at the same setting. If you regularly need to summarize a dense PDF or pull figures out of reports, this is the most practical upgrade.

On Terminal-Bench Science 0.1, which covers data analysis, simulations, and theorem proving, GPT-6.1 Sol more than doubles GPT-6 Sol's score at max effort for less than half the cost per task. At around $5.47 per task, it costs a fraction of GPT-6 Astra's roughly $23.80, though Astra remains OpenAI's strongest model on the hardest scientific work.

Across the tested reasoning settings, GPT-6.1 Sol's factual error rate stays within 1.9 percentage points of GPT-6 Astra's at under a fifth of the cost per task. At low effort, its error rate falls to 7.7% from 11.4% for GPT-6 Sol. Alignment improved as well: in OpenAI's adversarial tests, it tried to work around explicit restrictions in 23.5% of cases, down from 64.4% for GPT-6 Sol.

The table below compares GPT-6.1 Sol with the model it replaces and with OpenAI's flagship. Prices are OpenAI API rates per million tokens.

The pattern is clear. GPT-6.1 Sol closes most of the gap to Astra at the same price as GPT-6 Sol. The trade-off is speed: it generates text noticeably slower than GPT-6 Sol, so for quick everyday chats the older model can still feel more responsive.

GPT-6.1 Sol keeps GPT-6 Sol's API price of $2 per million input tokens and $10 per million output tokens — one-fifth of GPT-6 Astra's $10 and $50. Cached input drops to $0.10 per million tokens, half of GPT-6 Sol's rate, which matters for agents and long conversations that reuse context. OpenAI also offers an Ultrafast version that generates tokens up to eight times faster.

• ChatGPT: available in ChatGPT Work and Codex on Plus, Pro, Business, Enterprise, and Edu plans, but not yet in the regular Chat tab.
• Developers: available as gpt-6.1-sol in the OpenAI API and on Amazon Bedrock.
• Chat Smith: GPT-6.1 Sol is not integrated yet. You can use GPT-6 Sol today, with limited use on the free plan and higher usage limits on Chat Smith Pro at $9.99 per month or $39.99 per year.

GPT-6.1 Sol does not win every test. Independent results from Artificial Analysis largely back OpenAI's claims, but Anthropic and Google models still lead on some measures, and rival flagships such as Claude Opus 5.5 outscore it on certain agentic and scientific benchmarks at max effort.

It is also treated as a high-risk model in sensitive areas. Like Astra, OpenAI classifies GPT-6.1 Sol as having critical capability in cybersecurity and high capability in biology and chemistry, so it ships with the same safeguards — which can mean refusals or extra review steps on some security work.

Finally, it is slower than GPT-6 Sol and harder to reach. Until it arrives in ChatGPT's Chat tab and third-party apps, most everyday users will only meet it through Codex or developer tools. As always, test it against alternatives such as Claude Sonnet 5 on your real tasks before committing.

GPT-6.1 Sol is not integrated into Chat Smith yet, but its direct predecessor is. GPT-6 Sol is live in Chat Smith on Android, iOS, and the browser: free users get limited use, and Chat Smith Pro raises those limits for heavier work. Pick it from the model list and start chatting — no OpenAI account needed.

GPT-6 Sol shares GPT-6.1 Sol's price tier, 1M-token context window, and clearer GPT-6 answer style, and it is still the model behind GPT-6 in ChatGPT for paid users. For everyday writing, research, document work, and coding questions, it covers the same ground well. GPT-6.1 Sol's biggest gains show up mainly in long agentic tasks and scientific workflows.

Chat Smith also lets you compare GPT-6 Sol with DeepSeek V4 Pro, Gemini 3.5 Flash, and other leading models in the same app, so you can send one prompt to several models and keep the answer that holds up.

GPT-6.1 Sol is a same-price upgrade that changes the math on when you actually need Astra: near-flagship results in coding, document analysis, and professional work, better factuality and alignment, and cheaper caching, all at $2 and $10 per million tokens. Its main drawbacks are slower output and limited availability outside ChatGPT Work, Codex, and the API.

Until it reaches more apps, GPT-6 Sol remains the easiest way to use OpenAI's Sol line every day. You can try it now in Chat Smith, a free AI chatbot that puts it next to the latest models from OpenAI, Anthropic, Google, xAI, and DeepSeek.