GPT-6.1 SolNear-Astra coding at a fifth of the price
GPT-6.1 Sol is OpenAI's upgrade to GPT-6 Sol, released on 29 September 2026, one week after the model it replaces. OpenAI says it nearly matches GPT-6 Astra on agentic coding, computer use and professional work while keeping the same $2 per million input and $10 per million output tokens. It runs on Felo AI Search and the Felo API.
Live on the OpenAI API as gpt-6.1-sol since September 29, 2026 at $2 / $10 per million tokens — roughly one-fifth of GPT-6 Astra.
Released 29 September 2026 · Available now on Felo AI Search and the Felo API
Model card
GPT-6.1 Sol at a glance
- Model ID
- gpt-6.1-sol
- Released
- September 29, 2026
- Price vs. GPT-6 Astra
- one-fifth of Astra's $2 / $10
- Context window
- 1,050,000 tokens
- Position
- Upgrade to GPT-6 Sol, just below Astra
$2 / $10
Input price
Per million tokens, standard rate.
1.05M
Context window
1.05M tokens, 128,000 max output.
September 29, 2026
Released
One week after GPT-6 Sol.
gpt-6.1-sol
API model ID
The string to send in the model field.
Where 6.1 Sol sits in the GPT-6 family
GPT-6.1 Sol is not a new tier. It is the same slot as GPT-6 Sol, refreshed a week later with benchmarks that move it next to Astra while the price stays put.
An upgrade, not a replacement tier
OpenAI describes GPT-6.1 Sol as an upgrade to GPT-6 Sol that nearly matches GPT-6 Astra's intelligence on agentic coding, computer use and professional work. The model ID changes from gpt-6-sol to gpt-6.1-sol; the position in the family does not.
Astra performance, Sol pricing
GPT-6 Astra costs $10 input / $50 output per million tokens. GPT-6.1 Sol stays at $2 / $10, a fifth of the flagship, while OpenAI's own benchmarks put it in line with Astra on the work most teams actually run. The gap between the two tiers narrowed without the price moving.
Same API surface as GPT-6 Sol
GPT-6.1 Sol keeps the 1,050,000-token context window, the 128,000-token maximum output, the reasoning-effort scale and the tool set of GPT-6 Sol. Moving a workload onto it is a model-ID change, not a rewrite.
What the upgrade changes
Four measured differences between GPT-6.1 Sol and the model it replaced one week earlier.
Better on agentic coding
On DeepSWE 1.1, OpenAI's coding benchmark, GPT-6.1 Sol scores 6.4 percentage points higher than GPT-6 Sol, landing essentially level with GPT-6 Astra at one-fifth of the cost.
Stronger computer use
At maximum reasoning effort, GPT-6.1 Sol outperforms GPT-6 Sol on computer use by seven percentage points, at half the cost per task, and again performs in line with Astra.
Fewer factual errors
At low reasoning effort, the share of responses containing at least one factual error fell from 11.4% with GPT-6 Sol to 7.7% with GPT-6.1 Sol, a reduction of about 32% on OpenAI's deliberately difficult test set.
Better at respecting intent and safety
OpenAI says the new model is better at respecting user intent and safety constraints. In its tests it failed to disclose broken search tools in only 2.1% of cases, and agents built on it never tried to work around an automated safety reviewer's block.
GPT-6.1 Sol benchmarks and pricing
OpenAI published the comparison figures. They are reproduced here beside the models the page is compared with, with the source named.
Price per million tokens
Standard API rates in USD. GPT-6.1 Sol keeps GPT-6 Sol's price while moving toward Astra's benchmark scores.
Model
GPT-6.1 Sol
GPT-6 Sol
GPT-6 Astra
Input
$2
$2
$10
Output
$10
$10
$50
Cached input
$0.10
$0.20
$1
GPT-6.1 Sol against GPT-6 Sol and GPT-6 Astra
OpenAI's published scores on the three benchmarks it highlighted at launch. GPT-6.1 Sol closes most of the gap to Astra while staying at Sol's price.
Source: OpenAI launch figures, as reported by The New Stack, 29 September 2026.
Agentic coding and computer use
OpenAI's published comparison between GPT-6.1 Sol, GPT-6 Sol and GPT-6 Astra. Higher is better.
Benchmark
GPT-6.1 Sol
GPT-6 Sol
GPT-6 Astra
Coding
DeepSWE 1.1
75.0%
68.6%
75.0%
Computer use and documents
GDP.pdf
32.0%
28.0%
32.0%
Computer use (max effort)
32.0%
25.0%
32.0%
Price per million tokens
GPT-6.1 Sol costs a fifth of GPT-6 Astra on both input and output, and the same as the GPT-6 Sol it replaces.
Source: OpenAI API pricing, 30 September 2026.
Context window
GPT-6.1 Sol keeps the 1,050,000-token context window and 128,000-token maximum output of GPT-6 Sol.
Source: OpenAI model catalog, 30 September 2026.
Cost against GPT-6 Astra
GPT-6.1 Sol keeps GPT-6 Sol's $2 / $10 rate, a fifth of GPT-6 Astra's $10 / $50, while its benchmark scores move next to Astra.
Source: OpenAI API pricing, 30 September 2026.
Benchmark figures are OpenAI's own published comparisons. They are not independent measurements, and they do not represent typical use.
What is confirmed, and when
The dates and sources behind the claims on this page.
22 September 2026
OpenAIGPT-6 Sol ships
OpenAI releases GPT-6 Sol and GPT-6 Luna, priced at $2 / $10 and $0.10 / $0.50 per million tokens.
29 September 2026
OpenAIGPT-6.1 Sol ships
OpenAI releases GPT-6.1 Sol as an upgrade to GPT-6 Sol, one week after the original. It is available in the API and to Plus, Pro, Business, Enterprise and Edu users in ChatGPT Work and Codex, but not in Chat.
29 September 2026
ReportedBenchmarks published
OpenAI's launch figures put GPT-6.1 Sol 6.4 points above GPT-6 Sol on DeepSWE 1.1, seven points above it on computer use, and in line with GPT-6 Astra on both.
30 September 2026
OpenAIListed in the model catalog
OpenAI's model catalog lists gpt-6.1-sol with a 1.05M context window, 128K maximum output, an April 30, 2026 knowledge cutoff and $2 / $10 pricing.
Where a claim comes from a vendor's own benchmark, the source is named. Where it comes from reporting, the outlet and date are named. No figure on this page is an estimate.
GPT-6.1 Sol specifications
The published model card, in one place.
Reasoning effort
reasoning.effort supports low, medium (default), high, xhigh and max, so a routine step can run cheap and a hard one can think longer.
Price per million tokens
Input $2.00 · Cached input $0.10 · Cache writes $2.50
Output $10.00 · Batch and Flex half price · Fast mode 2x
Prompts above 272K input tokens are billed at 2x input and 1.5x output for the full request.
Context window
1,050,000 tokens · 128,000 maximum output
Knowledge cutoff 30 April 2026
Availability
API, ChatGPT Work and Codex for Plus, Pro, Business, Enterprise and Edu users. Not available in Chat at launch.
Replaces
GPT-6 Sol, which shipped one week earlier. The model ID changes from gpt-6-sol to gpt-6.1-sol; the tier does not.
Benchmarks
DeepSWE 1.1 75.0%, GDP.pdf 32.0%, computer use at maximum effort 32.0% — OpenAI launch figures.
Inputs and outputs
What GPT-6.1 Sol accepts and returns.
Text in, text out
The standard path on the Responses and Chat Completions APIs, with function calling, structured outputs and streaming.
Image input
Screenshots, diagrams and scanned documents go in directly. Text and image in, text out.
Long documents in one pass
A 1,050,000-token window takes whole repositories and document sets rather than excerpts, so the retrieval layer in front of the model can come out.
What to run on GPT-6.1 Sol
The jobs OpenAI built the model for, and where the price makes the difference.
Agentic coding
Feature work, code review, debugging and multi-step tasks across a real repository, where GPT-6.1 Sol now scores near Astra at a fifth of the price.
Computer use
Agents that drive a browser or desktop. The seven-point gain over GPT-6 Sol at maximum reasoning is the headline improvement.
Professional work at volume
Drafting, analysis and document work where near-Astra quality matters but Astra's $10 / $50 rate does not fit the budget.
Document and PDF analysis
Complex PDFs and long document sets, where GPT-6.1 Sol tops out around 32% on GDP.pdf against about 29% for Claude Opus 5.5.
High-volume batch work
Summarising, classifying and extracting across large document sets, where the per-token price decides whether the job is worth running at all.
Tool and API integration
Structured output at volume for pipelines, with web search, file search, code interpreter, hosted shell and MCP available on the Responses API.
Run the GPT-6 family on Felo
Start with GPT-6.1 Sol
The upgraded mid-tier model, live as gpt-6.1-sol since September 29, 2026: $2 in and $10 out per million tokens, with a 1,050,000-token context window and sources attached to every answer.
Open GPT-6.1 SolMove up to GPT-6 Astra
The flagship, for the hardest single-shot reasoning and computer use. GPT-6.1 Sol now covers most of that ground at a fifth of the price, so reserve Astra for the tasks that still need it.
Open GPT-6 AstraFrequently asked questions
GPT-6.1 Sol is OpenAI's upgrade to GPT-6 Sol, released on 29 September 2026, one week after the original. OpenAI says it nearly matches GPT-6 Astra on agentic coding, computer use and professional work while keeping GPT-6 Sol's $2 / $10 per million token price. It is live on the API as gpt-6.1-sol.
Try GPT-6.1 Sol on Felo
OpenAI's upgraded mid-tier model: near-Astra agentic coding and computer use at $2 in and $10 out per million tokens, with a 1,050,000-token context window.
Available on Felo AI Search and the Felo API