GPT 6.1 Sol.

OpenAI used DevDay 2026, on September 29, to make a pricing argument as much as a capability one. GPT-6.1 Sol, its upgraded mid-tier model, claims intelligence close to GPT-6 Astra — the flagship — on agentic coding, computer use and professional work, at one-fifth of Astra’s token prices. At $2 per million input tokens and $10 per million output, it is priced like a workhorse. One catch up front: at launch it does not appear in the regular ChatGPT chat window at all.

What Is GPT-6.1 Sol?

GPT-6.1 Sol is OpenAI’s mid-tier GPT-6 model, released as an upgrade to GPT-6 Sol. Developers call it in the API as gpt-6.1-sol. It is a reasoning model with a 1,050,000-token context window and a maximum output of 128,000 tokens, so one request can hold a large repository or document set. It accepts text and image input and returns text, with an April 30, 2026 knowledge cutoff.

Astra remains OpenAI’s answer for the very hardest problems. Sol is the model you run all day — inside coding agents, document pipelines and business automations, where cost per finished task decides whether the product makes money. It is also more careful than its predecessor: on deliberately difficult prompts, responses containing a factual error fell from 11.4% (GPT-6 Sol) to 7.7% at low reasoning effort.

Key Features

• Near-Astra capability: On DeepSWE v1.1 (complex software engineering in real codebases), GPT-6.1 Sol matches GPT-6 Astra at roughly one-fifth of the cost, and beats GPT-6 Sol’s best score by 6.4 points.
• Computer use: On OSWorld 2.0’s offline set, Sol improves on GPT-6 Sol by seven points and lands within 2.1 points of Astra — at roughly one-seventh of Astra’s cost per task.
• Very cheap cached input: Cached input costs $0.10 per million tokens, 95% below the standard input rate and half of GPT-6 Sol’s cached price. Agent loops that re-send the same context benefit most.
• Full tool support: Via the Responses API: web search, file search, image generation, code interpreter, hosted shell, apply patch, computer use and MCP, plus function calling and structured outputs.
• Speed tiers when you need them: Fast mode costs 2x standard pricing, Batch and Flex 50% less, and Ultrafast offers up to 8x faster token generation in Codex at 6x standard API prices.
• Strong document work: On GDP.pdf (answers from complex PDFs with tables, charts and fine print), Sol beats Claude Opus 5.5 with fallbacks at less than half the cost per task.

How to Use GPT-6.1 Sol

1. Check where you can reach it. GPT-6.1 Sol is available to Plus, Pro, Business, Enterprise and Edu users inside ChatGPT Work and Codex, and to developers through the API. It is not yet in regular Chat, and free plans are excluded.
2. For coding, open Codex. Hand it a real task with clear boundaries, and review the output like a pull request from a fast junior engineer.
3. For API work, call gpt-6.1-sol in the Responses API. Tool calling needs the Responses API (Chat Completions works only without tools); set reasoning.effort to suit the job.
4. Cache your repeated context. Cache stable system prompts, repo context and reference documents. At $0.10 per million cached tokens, uncached repetition is wasted money.
5. Watch the 272K line. Prompts over 272K input tokens are billed at 2x input and cache rates and 1.5x output for the whole request — splitting a giant request can be cheaper.

GPT-6.1 Sol Pricing

GPT-6.1 Sol is a Paid model with per-token API pricing and no free tier. There is no single starting price: cost depends on the route — subscription access via ChatGPT Work and Codex, or metered API usage:

Plan	Price	What you get
API (standard)	$2 in / $10 out	Per million tokens. Cached input $0.10 per million, cache writes $2.50. The headline rate applies to prompts up to 272K input tokens.
API Batch / Flex	$1 in / $5 out	Per million tokens — 50% below standard pricing for work that does not need an instant response.
ChatGPT plans	Plus and above	Access inside ChatGPT Work and Codex on Plus, Pro, Business, Enterprise and Edu plans. Not on Free or Go, and not yet in regular Chat.

OpenAI’s own figures put an average Terminal-Bench Science task at $5.47 on Sol, against $23.21 for Claude Opus 5.5 and $23.80 for Astra — Sol exists to make frontier-adjacent work cheap enough to run constantly.

Pros

• Near-flagship results on coding, computer use and document work at one-fifth of Astra’s token prices
• Cached input at $0.10 per million makes long agent loops genuinely affordable
• Huge 1.05M-token context window with 128K maximum output
• Full Responses API toolset, including computer use, code interpreter and MCP
• Measurably fewer factual errors than GPT-6 Sol on difficult prompts
• Flexible tiers — Batch/Flex for savings, Fast and Ultrafast for deadlines

Cons

• Not available in regular ChatGPT Chat at launch — only ChatGPT Work and Codex
• No free access anywhere: free plans and the API free tier are excluded
• Prompts above 272K input tokens cost double on input for the entire request
• Astra still beats it on the hardest scientific research, where OpenAI recommends Astra
• The “none” reasoning setting is gone — low is the floor

Who Should Use GPT-6.1 Sol?

Development teams building coding agents or running Codex all day are the clearest fit: the DeepSWE results and the cached-input price are aimed squarely at them. It also suits teams automating document-heavy workflows — contracts, filings, PDFs full of tables — where accuracy at low cost per task matters. If you mainly chat with AI in the standard ChatGPT window, this launch passes you by for now.

GPT-6.1 Sol Alternatives

The head-to-head comparison most teams will run is against Anthropic’s Claude Opus 5.5, which leads on several coding benchmarks but charges double Sol’s token price. Google’s Gemini 4 Argon is the third contender for long, complex professional work. All three sit in our AI tools directory.

Frequently Asked Questions

How much does GPT-6.1 Sol cost?

Standard API pricing: $2 per million input tokens, $0.10 cached input, $10 output. Batch and Flex cost half; Fast mode costs double.

Is GPT-6.1 Sol free?

No. It requires a paid route — a Plus, Pro, Business, Enterprise or Edu plan for ChatGPT Work and Codex, or paid API usage. The API free tier does not support it.

What is the GPT-6.1 Sol context window?

1,050,000 tokens, with up to 128,000 output tokens and an April 30, 2026 knowledge cutoff. Prompts over 272K input tokens cost more.

GPT-6.1 Sol vs GPT-6 Astra — which should I choose?

Sol for everyday coding, computer use and professional work — near-Astra results at a fifth of the token price. Astra for the most difficult scientific research, where it still scores highest.

Final Verdict

GPT-6.1 Sol squeezes the middle of the market: most of Astra’s practical ability, priced for volume. The benchmarks are strong, the caching price is aggressive, and the accuracy gains matter for unsupervised agents. The frustrations are access — no regular Chat, no free tier — and a long-context surcharge that punishes lazy prompt design. If you build with the API or live in Codex, trial it first. Visit OpenAI’s official GPT-6.1 Sol announcement for the full launch details.