GPT-6.1 Sol: OpenAI's Near-Astra Model at a Fifth of the Price (2026)

The short answer

GPT-6.1 Sol is a mid-tier workhorse that closes most of the gap between the everyday GPT-6 Sol and the top-of-the-line GPT-6 Astra. OpenAI positions it as the model to reach for when you want near-Astra quality on hard coding and multi-step business tasks, but don't want to pay Astra prices^[1].

The headline number is the price. Standard API rates are $2.00/M input and $10.00/M output — versus $10.00/M and $50.00/M for GPT-6 Astra^[1]. Cached input drops to just $0.10/M, which OpenAI says is 95% less than its standard input price and half of what GPT-6 Sol charged for cached input^[1].

What's new vs GPT-6 Sol

GPT-6.1 Sol is an upgrade to GPT-6 Sol, not a brand-new architecture. The gains are concentrated in the areas where Sol already played — writing and debugging code, reading dense documents, and executing multi-step business workflows^[1].

OpenAI's published evaluations show the improvement is real rather than cosmetic. On DeepSWE v1.1, which tests agents on real, long-horizon software-engineering tasks, GPT-6.1 Sol matches GPT-6 Astra's score at roughly one-fifth the cost, and beats GPT-6 Sol's best result by 6.4 percentage points at lower reasoning effort^[1]. On AutomationBench, Zapier's benchmark for end-to-end business workflows, it scores 2.2 points above Claude Opus 5.5 at medium effort for roughly a third of the cost^[1].

Where it wins on benchmarks

• Coding (DeepSWE v1.1): matches GPT-6 Astra at ~1/5 the cost; up 6.4 points on GPT-6 Sol's best^[1].
• Professional documents (GDP.pdf): scores above Claude Opus 5.5 with fallbacks at less than half the cost per task^[1].
• Automation (AutomationBench): +2.2 points vs Opus 5.5 at ~1/3 cost; +4.8 points vs GPT-6 Sol^[1].
• Computer use (OSWorld 2.0 offline): +7 points over GPT-6 Sol at maximum effort; within 2.1 points of Astra at ~1/7 the cost^[1].
• Science (Terminal-Bench Science 0.1): more than doubles GPT-6 Sol's score, averaging $5.47 per task versus $23.21 for Opus 5.5 and $23.80 for Astra^[1].
• Factuality: at low reasoning effort, factual-error rate falls from 11.4% to 7.7% (about a 32% reduction)^[1].

Pricing and availability

GPT-6.1 Sol is available starting today to all Plus, Pro, Business, Enterprise, and Edu users inside ChatGPT Work and Codex^[1]. It is not yet available in ChatGPT Chat, so consumers on the standard assistant won't see it immediately^[1]. Developers can call it through the API under the model ID gpt-6.1-sol^[1].

An Ultrafast variant is coming within days, promising up to 8x faster token generation compared to standard speed in Codex^[1][3]. Safety details are published in the GPT-6.1 Sol system card addendum^[2].

What it means for you

• Developers get the biggest win. If you build agents that reuse context across requests, the $0.10/M cached input price is a meaningful cost cut over both GPT-6 Sol and Astra^[1].
• Business and Work users can upgrade for free. The model simply becomes available in ChatGPT Work and Codex on existing paid plans — no price change required^[1].
• Casual ChatGPT users should wait. GPT-6.1 Sol isn't in the standard Chat experience yet, so there's nothing to switch for if you only use the consumer assistant^[1].

Bottom line

Verdict: GPT-6.1 Sol is the clearest "price-performance" release from OpenAI in a while — near-Astra capability on the tasks that matter to builders, at a fraction of the cost. It's aimed squarely at developers and professionals running agents at volume, not at the casual ChatGPT user.

Frequently asked questions

Is GPT-6.1 Sol the same as GPT-6 Sol?

No. GPT-6.1 Sol is a newer upgrade to GPT-6 Sol with higher benchmark scores and lower cached-input pricing^[1].

How much does GPT-6.1 Sol cost?

$2 per million input tokens, $0.10 per million cached input tokens, and $10 per million output tokens via the API^[1].

Where can I use GPT-6.1 Sol?

In ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu plans, and via the API as gpt-6.1-sol^[1]. It is not yet available in ChatGPT Chat.