OpenAI released GPT-6.1 Sol at DevDay on September 29, 2026. It’s an upgrade to GPT-6 Sol that OpenAI says comes close to GPT-6 Astra on coding, computer use and professional work, at a fifth of Astra’s price. It keeps GPT-6 Sol’s list price in the API and halves the cost of cached input. In ChatGPT, it only runs in Work and Codex for now.
The short version
- API prices are $2 per million input tokens, $0.10 cached and $10 output: the same list price as GPT-6 Sol, with cached input halved.
- OpenAI says it matches GPT-6 Astra on the DeepSWE v1.1 coding benchmark at about a fifth of the cost.
- It’s in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise and Edu, but not in regular chat, and not on Free or Go at launch.
Where you can use GPT-6.1 Sol
The launch post says GPT-6.1 Sol is available “to all Plus, Pro, Business, Enterprise, and Edu users in ChatGPT Work and Codex” and is “not yet available in Chat”. Developers reach it through the API as gpt-6.1-sol. OpenAI’s DevDay recap lists it for API, Plus, Pro, Business, Enterprise and Edu users without mentioning the Chat limit.
GPT-6.1 Sol availability by product and plan
| Where | Who gets it |
|---|---|
| Codex (desktop app and CLI) | Plus, Pro, Business, Enterprise, Edu |
| ChatGPT Work (web and mobile) | Plus, Pro, Business, Enterprise, Edu |
| Regular ChatGPT chat | Not yet available |
| OpenAI API | Usage tiers 1 to 5 (not the free tier) |
| Free and Go plans | Not included at launch |
On Enterprise and Edu, the model stays off until an administrator enables it, according to OpenAI’s model documentation. Standard and Fast speeds are available now. OpenAI’s Thibault Sottiaux (Tibo) described it as “unbelievably efficient” and “included in all paid plans”, though Go, a paid plan, is not on the launch list.
GPT-6.1 Sol API pricing compared with GPT-6 Sol and Astra
OpenAI’s headline is price against Astra: one-fifth of Astra’s standard input and output rates. Against GPT-6 Sol, which it upgrades, the list price hasn’t moved.
Standard API prices per million tokens
| Model | Input | Cached input | Output |
|---|---|---|---|
| GPT-6 Astra | $10.00 | $1.00 | $50.00 |
| GPT-6.1 Sol | $2.00 | $0.10 | $10.00 |
| GPT-6 Sol | $2.00 | $0.20 | $10.00 |
| GPT-6 Luna | $0.10 | $0.01 | $0.50 |
These are short-context rates for prompts up to 272K input tokens, from OpenAI’s API pricing page and, for GPT-6 Sol, its September 22 API changelog. Longer prompts cost more. OpenAI says the cheaper cached input gives developers more room to run agents that reuse context across requests, since cached tokens cost 95% less than fresh input.
How GPT-6.1 Sol scores against Astra and Opus 5.5
OpenAI ran the GPT evaluations in its research environment or through its API, and took the competitor results from publicly available reports.
OpenAI’s GPT-6.1 Sol benchmark results
| Benchmark | What OpenAI reports |
|---|---|
| DeepSWE v1.1 (coding) | Matches GPT-6 Astra at roughly a fifth of the cost; 6.4 points above GPT-6 Sol’s best |
| GDP.pdf (documents) | Above Opus 5.5 with fallbacks at less than half the cost per task |
| AutomationBench (workflows) | 2.2 points above Opus 5.5 at medium effort, at roughly a third of the cost |
| OSWorld 2.0 offline set (computer use) | At max effort, 7 points above GPT-6 Sol; within 2.1 points of Astra at about a seventh of the cost |
| Terminal-Bench Science 0.1 | At max effort, $5.47 per task on average vs $23.21 for Opus 5.5 and $23.80 for Astra |
GPT-6 Astra still has the top Terminal-Bench Science score among the models tested, 68.1%, and OpenAI says it should be used for the most difficult scientific research tasks. GPT-6.1 Sol’s biggest factual-accuracy gain over GPT-6 Sol came at low reasoning effort, where the share of answers with at least one factual error fell from 11.4% to 7.7%. OpenAI measured that on deliberately difficult prompts where users had flagged an earlier model’s mistake.
GPT-6.1 Sol context window and specs
OpenAI’s model page lists a 1,050,000-token context window, up to 128,000 output tokens and a knowledge cutoff of April 30, 2026. Reasoning effort runs from low to max. In the Responses API, GPT-6.1 Sol can also hand parts of a request to subagents, a feature OpenAI calls multi-agent, which is in beta.
Ultrafast, the premium speed tier, isn’t available for GPT-6.1 Sol yet. The launch post says it arrives “in the coming days” with up to 8x faster generation in Codex. For now, Ultrafast in the API is broadly available only for GPT-6 Astra, and in ChatGPT Work and Codex it’s limited to Pro 500 and eligible Enterprise and Edu plans.
How GPT-6.1 Sol fits OpenAI’s GPT-6 lineup
GPT-6.1 Sol is the third GPT-6 release in four weeks.
OpenAI’s guidance is to use GPT-6.1 Sol for complex coding and long-running work, and Luna for focused, repeatable tasks. The older GPT-5.6 models stay available during the rollout.
FAQs
Can I use GPT-6.1 Sol in the normal ChatGPT chat?
Not yet. It’s limited to ChatGPT Work and Codex on paid plans from Plus up.
Is GPT-6.1 Sol on ChatGPT Go?
No. OpenAI’s model documentation says Free and Go aren’t included at launch. In Codex, Free and Go plans get GPT-6 Luna at Standard speed in the desktop app, subject to rollout.
What’s the API model name?
gpt-6.1-sol, available in the Responses and Chat Completions APIs. Tool calling needs the Responses API.
Related
Leave a Reply