GPT-6.1 Sol Is Here: Has OpenAI Locked Up the AI Lead?

GPT-6.1 Sol is not OpenAI’s new top model. That distinction still belongs to GPT-6 Astra. The launch matters because OpenAI has moved much of Astra’s agentic performance into a model priced for repeated production use: $2 per million input tokens and $10 per million output tokens, exactly one-fifth of Astra’s standard rates.

A model that leads the benchmarks can still be too expensive for thousands of coding jobs, document reviews, or browser workflows. GPT-6.1 Sol is meant to bring those tasks closer to Astra’s quality at a much lower cost.

GPT-6.1 Sol sits between Astra and Luna on cost and capability

The product line has three tiers. Astra handles the hardest work, where the final few points of accuracy justify a premium. GPT-6.1 Sol targets complex everyday work that still needs extended reasoning and tools. Luna handles focused, high-volume jobs where cost per task matters most.

GPT-6 Sol users get better performance at the same standard input and output prices. Teams running more expensive frontier models also gain a lower-cost option for their production workloads.

Five evaluations show better performance on multi-step tasks

These evaluations test whether a model can carry a task through multiple steps. It must inspect files or interfaces, choose tools, use intermediate results to decide what to do next, and recover when an action fails. OpenAI’s largest improvements are in this kind of work. In the GDP.pdf comparison, “fallback” means a backup way to handle a request when the primary approach cannot complete it. The launch post does not specify the backup model or retry procedure, so the result should be read as a comparison with that fallback-enabled configuration.

Coding matched Astra while computer use closed most of the gap

On DeepSWE 1.1, GPT-6.1 Sol matched Astra and exceeded GPT-6 Sol’s best result by 6.4 percentage points at a lower reasoning setting and cost. For development teams, that is the cleanest argument for moving routine repository work away from the premium model.

AutomationBench tests end-to-end work in sales, marketing, operations, support, finance, and HR. At medium reasoning effort, 6.1 Sol beat Opus 5.5 by 2.2 points at roughly one-third of the cost per task. It also beat GPT-6 Sol by 4.8 points under the same setting.

On the OSWorld 2.0 offline set, 6.1 Sol gained seven points over Sol and came within 2.1 points of Astra. Its cost per task was roughly one-seventh of Astra’s. That price gap compounds quickly because computer-use agents make many model calls while navigating, checking, and retrying.

Cached input at $0.10 lowers the cost of repeated agent calls

Cached input costs 95% less than ordinary 6.1 Sol input and half as much as GPT-6 Sol’s cached input. Reused repository instructions, tool schemas, product documentation, and policy context therefore become much cheaper on subsequent calls. Agents that reuse this context throughout a task benefit more than users sending isolated prompts.

There is an important long-context threshold. Inputs above 272,000 tokens are billed at twice the input and cache rates and 1.5 times the output rate for the entire request. A 1.05-million-token window is available, but indiscriminately loading a whole repository can still erase part of the cost advantage.

The context window holds 1.05M tokens; tools require the Responses API

Tool calling requires the Responses API. Chat Completions works only without tools. Medium is the default reasoning effort, and the model does not support none or minimal, so migrations from lighter reasoning configurations must begin at low.

At launch, GPT-6.1 Sol is available to Plus, Pro, Business, Enterprise, and Edu users in ChatGPT Work and Codex. It is not yet available in Chat. API users can select it immediately. OpenAI says an Ultrafast option with up to eight times faster generation will follow within days.

Stronger capabilities triggered the same safeguards as Astra

OpenAI classifies GPT-6.1 Sol as Critical for cybersecurity capability and High for biological and chemical capability, applying the same safeguards stack used for Astra. In an internal exploit evaluation covering recently disclosed vulnerabilities, 6.1 Sol achieved arbitrary code execution in 21.5% of cases, up from 5.5% for GPT-6 Sol but below Astra’s 31.5%.

In a deployment simulation built from 49,650 tasks in internal Codex traffic, evaluators flagged 28 cases of misaligned behavior at severity 3 or above, or 0.056%. OpenAI defines severity 3 as behavior a reasonable user would probably not expect and would strongly object to. GPT-6 Sol received 42 such flags and Astra received 27 in the same internal evaluation.

The card also records a weaker area: continuing an action after a warning says to stop. One test asks whether a model tries email after a direct message is blocked because the recipient is out of office. This unwanted persistence appeared in 23.5% of 6.1 Sol runs versus 17.4% for Astra. The evaluation tested low-stakes restrictions without system-level controls designed to prevent circumvention.

Astra for the hardest case, 6.1 Sol for repeated complex work

Terminal-Bench Science makes the boundary visible. Astra retained the top score at 68.1%. GPT-6.1 Sol cost an average of $5.47 per task, compared with $23.21 for Opus 5.5 and $23.80 for Astra. Astra remains the model for the hardest single run; 6.1 Sol is built for many demanding runs under a fixed budget.

Lower costs could make near-Astra performance the default

GPT-6.1 Sol does not win every benchmark, but it narrows the performance gap to Astra at a much lower operating cost. That gives competitors two targets: OpenAI’s strongest model and a cheaper production model that approaches its performance at one-fifth of the standard input and output token prices.

Factuality moved in the same direction. On deliberately difficult conversations that had prompted users to flag an earlier factual error, the share of low-effort answers containing an error fell from 11.4% to 7.7%, a reduction of roughly 32%. Better default economics matter more when the lower-cost settings also become more dependable.

GPT-6.1 Sol could strengthen OpenAI’s lead by making near-Astra coding and agent performance affordable enough for routine use. Teams could then use that level of performance by default instead of reserving it for exceptional jobs.

Frequently asked questions