OpenAI has launched GPT-6.1 Sol, an upgrade to GPT-6 Sol. According to the official announcement, the new model approaches GPT-6 Astra on agentic work such as coding, computer use, and professional tasks, at about one-fifth of Astra’s standard input and output token prices.
For current plans, prices and every Maitz article on ChatGPT, see our ChatGPT hub.
The standout cost figure: cached input is $0.10 per million tokens. That is about 95% off standard input pricing, and 50% less than the previous Sol cached rate. OpenAI frames this as more headroom for developers who run agents that reuse the same context across requests.
What improved versus Sol
The post cites several internal evaluations. On DeepSWE v1.1 (long software-engineering tasks in real codebases), GPT-6.1 Sol matches Astra at roughly one-fifth the cost, and beats Sol’s best score by 6.4 percentage points at a lower reasoning effort and cost.
On GDP.pdf (professional questions over complex PDFs with tables, charts, and fine print), it scores higher than Opus 5.5 with fallbacks at less than half the cost per task, and approaches Astra at about one-fifth the cost. On AutomationBench (multi-step business workflows with dozens of tools), it scores 2.2 points above Opus 5.5 at medium reasoning effort at about a third of the cost, and 4.8 points above Sol at the same setting.
Computer use, science, and factuality
On OSWorld 2.0’s offline set, GPT-6.1 Sol beats Sol by seven percentage points at maximum reasoning effort for less than half the cost, and comes within 2.1 points of Astra at about one-seventh the cost per task.
On Terminal-Bench Science 0.1 (data analysis, simulation, theorem proving), it more than doubles Sol’s score at maximum effort for less than half the cost. OpenAI notes that Astra still leads that suite at 68.1% and remains the pick for the hardest scientific research. On factuality at low reasoning effort, the share of answers with a factual error fell from 11.4% to 7.7% versus Sol (about a 32% reduction).
Pricing and availability
- API id: gpt-6.1-sol
- Prices: $2 per million input tokens, $0.10 per million cached input, $10 per million output
- Available to Plus, Pro, Business, Enterprise, and Edu users in ChatGPT Work and Codex
- Not yet available in Chat
- In the coming days, GPT-6.1 Sol Ultrafast is planned, with up to 8x faster token generation in Codex
Alignment and safety
OpenAI reports substantial alignment gains over Sol, bringing the model closer to Astra. Per the post, it is more transparent about its limits and more reliable at respecting user intent and safety constraints. Full details are in the GPT-6.1 Sol system card addendum.
How to read this
This is a stronger mid-tier model at a much lower price than Astra, not a replacement for Astra on the hardest work. For more OpenAI coverage on this site, see our piece on OpenAI’s Dots, compare with Google’s Gemini 4 Argon launch, and follow updates in AI News.
Sources
- OpenAI, Introducing GPT-6.1 Sol: openai.com/index/introducing-gpt-6-1-sol/
