GPT Sol 6.1: What OpenAI's GPT-6.1 Sol Is and How to Use It
Updated 2026-10-04
"GPT Sol 6.1" (also searched as "sol 6.1" or "6.1 sol") is GPT-6.1 Sol, an OpenAI model announced at OpenAI DevDay in San Francisco on September 29, 2026. It is an upgrade to GPT-6 Sol that sits below the flagship GPT-6 Astra: OpenAI positions it as near-Astra intelligence for agentic coding, computer use and professional work at roughly a fifth of Astra's standard API price. It is available in the API as gpt-6.1-sol, in Codex, and in ChatGPT Work for paid plans.
GPT-6.1 Sol key facts
| GPT-6.1 Sol | |
|---|---|
| Developer | OpenAI |
| Announced | September 29, 2026 (DevDay) |
| API model ID | gpt-6.1-sol |
| Price (standard API) | $2 input / $10 output per 1M tokens; cached input $0.10 per 1M |
| Context window | 1,050,000 tokens, up to 128,000 output tokens |
| Knowledge cutoff | April 30, 2026 |
| Positioned for | Agentic coding, computer use, document-heavy professional work |
These figures come from launch coverage of OpenAI's DevDay announcement and the model's API listings; OpenAI's own model page is the final authority if anything changes.
What "Sol" means in OpenAI's GPT-6 lineup
The GPT-6 generation is split into named tiers rather than "mini" and "pro" suffixes:
- GPT-6 Astra – the flagship for long, demanding end-to-end work (listed at $10 input / $50 output per 1M tokens).
- GPT-6 Sol – the cost-effective high-end tier for professional and agentic coding work ($2 / $10).
- GPT-6 Luna – the fast, low-cost tier for high-volume, latency-sensitive jobs such as chat and classification ($0.10 / $0.50).
GPT-6.1 Sol is a point release of the Sol tier. Standard input and output prices stayed the same as GPT-6 Sol, while cached input was halved from $0.20 to $0.10 per 1M tokens. So "6.1" is the version and "Sol" is the tier – which is why searches mix up "gpt sol 6.1", "sol 6.1" and "6.1 sol".
What changed compared with GPT-6 Sol and GPT-6 Astra
OpenAI's launch claims focus on three areas:
- Coding and debugging. OpenAI reports improvements over GPT-6 Sol on its coding evaluations and says 6.1 Sol approaches GPT-6 Astra on several of them.
- Cost per task. On OpenAI's own Terminal-Bench Science comparison, the reported cost was $5.47 per task for GPT-6.1 Sol versus $23.80 for Astra.
- Factual accuracy. OpenAI reported that the share of responses with factual errors on difficult prompts (low-reasoning setting) fell from 11.4% to 7.7%.
All of these are vendor-reported numbers from the launch. Independent benchmarks take a few weeks to appear, so treat them as a starting point and test on your own tasks.
How to access GPT-6.1 Sol
In Codex
The refreshed Codex CLI defaults to GPT-6.1 Sol from version 0.159.1. Update the CLI (npm install -g @openai/codex or your package manager), run codex, and sign in with a Plus, Pro, Business, Enterprise or Edu plan. If you are deciding between Codex and Anthropic's agent, see Claude Code vs Codex.
In ChatGPT
At launch GPT-6.1 Sol was offered in ChatGPT Work for Plus, Pro, Business, Enterprise and Edu users. Launch reports said it was not yet in the regular ChatGPT chat model picker and not available on the Free or Go plans. If you don't see it, that is the most likely reason.
Through the API
- Make sure your OpenAI organization is on a paid usage tier (launch coverage lists tiers 1–5; the free tier is excluded).
- Call the model with the ID
gpt-6.1-sol(a dated snapshot,gpt-6.1-sol-20260929, is also listed by API routers). - Use cached prompts for long, repeated context – cached input is $0.10 per 1M tokens.
- Use the Batch API for non-urgent jobs; batch pricing is listed at a 50% discount.
The model is also listed on Azure OpenAI and on routers such as OpenRouter.
Estimating your cost
Cost = input tokens ÷ 1,000,000 × $2 + output tokens ÷ 1,000,000 × $10
Example: a coding task that sends 200,000 input tokens and gets 20,000 output tokens back costs about $0.40 + $0.20 = $0.60 at standard rates, before caching or batch discounts. Agent runs often make many turns, so measure the whole task, not a single call. Reasoning tokens are billed as output, so higher reasoning effort raises the bill.
Should you switch to GPT-6.1 Sol?
- You use GPT-6 Sol today: yes, test it – same standard price, cheaper cached input, and claimed coding gains.
- You use GPT-6 Astra for coding agents: run both on a sample of real tasks; if 6.1 Sol passes, the price gap is about 5x.
- You need cheap, high-volume chat or classification: GPT-6 Luna is the cheaper tier.
- You compare across vendors: see our notes on Claude Opus vs Sonnet and Gemini for coding.
FAQ
Is GPT Sol 6.1 an official OpenAI model?
Yes. GPT-6.1 Sol was announced by OpenAI at DevDay on September 29, 2026, and is available under the API model ID gpt-6.1-sol. "GPT Sol 6.1", "sol 6.1" and "6.1 sol" are just different word orders of the same name.
How much does GPT-6.1 Sol cost?
Standard API pricing is $2 per million input tokens and $10 per million output tokens, with cached input at $0.10 per million. In ChatGPT and Codex it is included in eligible paid plans rather than billed per token.
Is GPT-6.1 Sol free to use?
Not at launch. It was not offered on the ChatGPT Free or Go plans, and API access requires a paid usage tier. Codex access requires a Plus plan or higher, or an API key.
Why is GPT-6.1 Sol not showing in my model list?
At launch it appeared in ChatGPT Work and Codex, not in the regular ChatGPT chat picker. In the API, check that your organization is on a paid tier and that you use the exact ID gpt-6.1-sol. Rollouts can also take time across accounts.
What is the context window of GPT-6.1 Sol?
The model is listed with a 1,050,000-token context window and up to 128,000 output tokens, with a knowledge cutoff of April 30, 2026.
GPT-6.1 Sol vs GPT-6 Astra – which is better?
Astra remains OpenAI's flagship for the longest, hardest tasks. GPT-6.1 Sol is positioned as close to Astra on coding and agent work at about one fifth of the standard token price. For most coding workloads it is worth testing 6.1 Sol first and escalating to Astra only when it fails.