Announced September 29, 2026 at DevDay, GPT-6.1 Sol is an upgrade to GPT-6 Sol that OpenAI says approaches GPT-6 Astra on agentic coding, computer use, and professional work — while keeping Sol's low token price.
2026-09-29 · by Moe Ameen
On September 29, 2026, at its DevDay event, OpenAI introduced GPT-6.1 Sol, an upgrade to the GPT-6 Sol model it had released just a week earlier on September 22. The pitch is a value one: OpenAI says GPT-6.1 Sol reaches nearly the same level of intelligence as its flagship GPT-6 Astra on agentic coding, computer use, and professional work, while costing about one-fifth as much for standard input and output tokens.
The pricing lands GPT-6.1 Sol at $2 per million input tokens and $10 per million output tokens — the same standard rates as GPT-6 Sol — with cached input cut in half to $0.10 per million (from $0.20). Against GPT-6 Astra's $10 input and $50 output, those standard prices are exactly one-fifth. OpenAI's argument is that you no longer need to pay flagship rates for most agentic and professional work. On its internal factuality measure the model's error rate lands within about 1.9% of Astra across reasoning settings, and at low reasoning effort OpenAI reported the error rate dropping from 11.4% on GPT-6 Sol to 7.7% on GPT-6.1 Sol. It also cited benchmark parity with Astra on a software-engineering evaluation and a near-match on a computer-use test at maximum reasoning.
GPT-6.1 Sol is available through the API as `gpt-6.1-sol` and is rolling out to ChatGPT Plus, Pro, Business, Enterprise, and Edu users inside ChatGPT Work and Codex — not yet in the standard Chat surface. OpenAI said an "Ultrafast" variant that generates tokens much faster, at a higher price, is due in the coming days. Notably, the company said it scrapped a planned GPT-6.1 Astra over safety concerns during testing. Treat the exact benchmark figures, pricing, and rollout timing as a launch-window snapshot and confirm them against OpenAI's own pages before relying on them.
The useful way to read this launch is not "a smarter model" but "frontier-tier drafting at a commodity price." That lands squarely at the copy layer of a content pipeline, which is exactly where [Kompozy](/) already puts an OpenAI or Claude model to work: drafting captions, text posts, blog articles, and newsletter bodies governed by a [Persona Brief](/glossary/persona-brief) so the words read like you and respect your banned-word list. A model whose factual-error rate now approaches the flagship's means fewer weak drafts reaching Kompozy's per-post review gate — and on the Founding tier, where you bring your own key, GPT-6.1 Sol's low token price directly cuts what it costs to generate copy in bulk. The practical move today is to let the cheaper, sharper model do the drafting and keep your attention on approving output.
But a near-Astra model still just returns text — it ships nothing. Kompozy is the layer that turns that text into finished, published content: from one idea it generates [Persona Shorts](/glossary/persona-shorts), [Clipped Shorts](/glossary/clipped-short), brand-exact [Carousel Posts](/glossary/hyperframes), Quote Graphics, and full blogs and newsletters across [18 formats](/glossary/output-buckets), then schedules and publishes them across the eight social platforms plus blog and email on [Autopilot](/glossary/autopilot). When the model underneath gets this much cheaper and more reliable, the economics of the whole pipeline improve — but the thing that decides whether you actually post consistently is still the engine that produces and distributes the content, not the tokens beneath it. For the model this upgrades, see our coverage of [GPT-6 Sol and Luna](/news/openai-gpt-6-sol-and-luna-launch); for the flagship it chases, [GPT-6 Astra](/news/openai-gpt-6-astra-launch).
GPT-6.1 Sol is an upgrade to OpenAI's GPT-6 Sol model, announced September 29, 2026 at DevDay. OpenAI says it nearly matches the flagship GPT-6 Astra on agentic coding, computer use, and professional work while costing about one-fifth as much for standard tokens. It is a text and reasoning model, aimed first at builders in Codex and ChatGPT Work.
OpenAI listed GPT-6.1 Sol at $2 per million input tokens and $10 per million output tokens — the same standard rate as GPT-6 Sol — with cached input cut to $0.10 per million. Against GPT-6 Astra's $10 input and $50 output, those standard prices are exactly one-fifth. Confirm current figures on OpenAI's own pricing pages.
It keeps GPT-6 Sol's standard token price but improves accuracy and agentic performance to near-flagship levels, halves cached-input cost, and OpenAI reports a lower factual-error rate. At low reasoning effort the error rate dropped from 11.4% on GPT-6 Sol to 7.7% on GPT-6.1 Sol, and it matched flagship Astra on a software-engineering benchmark.
No. GPT-6.1 Sol is a text and reasoning model — it drafts, codes, and analyzes, but generates no video or images and publishes to no platform. To turn its drafts into finished, scheduled posts across platforms, you pair it with a content engine like Kompozy that generates video, carousels, and posts from one idea and publishes across nine platforms.