GPT-Image 2.6 Flare & Sunburst are liveImages from $0.025 each

GPT-6.1 Sol API

NewLLM

The GPT-6.1 Sol API is OpenAI's frontier GPT-6.1 reasoning route, with 1M context, image input, and tool use.

  • Input · <=272K$0.225 USDOfficial price$2 USD
  • Output · <=272K$1.464 USDOfficial price$10 USD
  • Input · >272K$0.45 USDOfficial price$4 USD
  • Output · >272K$2.929 USDOfficial price$15 USD
  • Savings91%
  • Context1M
  • View tool pricing and billing rules

What are we making today?

Bring an idea, a question, or a first draft.

Enter to send · Shift + Enter for a new line

01

What this model is for

The GPT-6.1 Sol API is OpenAI's frontier GPT-6.1 reasoning route, with 1M context, image input, and tool use.

GPT-6.1 Sol is OpenAI's upper-tier model in the GPT-6.1 line, released on September 29, 2026 under the id gpt-6.1-sol. OpenAI describes it as the choice for complex coding, computer use and professional work, one step below the flagship GPT-6 Astra. The catalog lists a 1M context window, image input, tool use and a 128,000-token output ceiling shared with Astra and GPT-6 Sol. Its standard input and output rates sit at the older Sol's level, while cache reads are listed lower.

Pick GPT-6.1 Sol when the integration already speaks the OpenAI API and the reasoning budget must stay visible. It accepts low, medium, high, xhigh and max effort, but rejects none and minimal, and it calls tools only through the Responses API. GPT-6.1 Terra covers high-volume forms, drawings and desktop automation at lower rates, and GPT-6 Astra stays for the most costly investigations. The knowledge cutoff is April 30, 2026, about a month earlier than Terra's.

Choose GPT-6.1 Sol for the hardest GPT-6.1 reasoning and agent work; its cache-hit rate is half of GPT-6 Sol's.

Best-fit coding workloads

Cross-file refactoring

Change code that spans packages: put the repository map, the failing tests and the migration notes in one 1M-token request, then ask for a patch plan before edits. Run the project's own test command after every patch, and raise effort only for the files that still fail.

Computer-use agents

Drive desktop software through tool calls: screenshot in, a short plan out, then one action per turn. Sol's tool use runs on the Responses API, so a computer-use loop keeps one conversation and one permission set. Log each action so a failed step can be replayed without repeating the whole run.

Two-pass code review

Run a first pass at low effort to catch naming, dead code and missing tests, then repeat at high or max on the files that matter. The explicit effort dial makes the split cheap to control. Ask for file and line references so a reviewer can check every finding.

02

Hands-on notes

Checked on 2026-10-27 through the OmniaKey API with one standard coding prompt (median of 3 runs) and three capability probes.

Output tokens for our standard coding prompt, reasoning included
197
Cost of that output at current prices
$0.00029
JSON mode
Passed
Tool calling
Passed
Image input
Passed

03

Technical specification

Catalog facts and integration details verified for this route.

Model ID
gpt-6.1-sol
Context
1M
Inputs and tools
Vision · Tool use
API endpoint
/v1/chat/completions
Cache hit / 1M
$0.0093 USD
Coding clients
Codex · Cursor · Cline · Aider

Limits to plan around

  • Effort levels none and minimal are rejected, so an existing agent that disables reasoning needs a new setting; temperature and top_p must also be dropped while reasoning is on.
  • Tool calling requires the Responses API. Chat Completions still serves plain text requests, but a function-calling agent built on it needs a real interface change, not just a new model id.
  • Cached input counts toward the prompt, and a cache write costs more than plain input, so a prefix that keeps changing can cost more than not caching at all.
  • The knowledge cutoff is April 30, 2026. Library releases and API changes after that date have to arrive through the prompt or through retrieval.

04

Compare nearby models

Compare workload fit and current listed token rates before choosing a route.

ModelBest forContextInputOutput
GPT-6.1 SolThis page
gpt-6.1-sol
The default GPT-6.1 pick for hard coding, review and tool-driven agent work.1M
$0.225 USD
Official price $2 USD
$1.464 USD
Official price $10 USD
GPT-6.1 Terra
gpt-6.1-terra
Better for high-volume forms, drawings and desktop work, where Terra's lower rates and later cutoff win.1M
$0.1125 USD
Official price $1 USD
$0.675 USD
Official price $5 USD
GPT-6 Astra
gpt-6-astra
Keep Astra for the most costly investigations, where a failed attempt justifies the flagship tier.1M
$0.9 USD
Official price $10 USD
$4.5 USD
Official price $50 USD
GPT-6 Sol
gpt-6-sol
Older version with the same input and output rates but a higher cache-read rate; keep it for pinned agents.1M
$0.225 USD
Official price $2 USD
$1.464 USD
Official price $10 USD

05

SDK quickstarts

Use one base URL and model ID from cURL, Python, or TypeScript.

Streaming chat completion
curl https://api.omniakey.com/v1/chat/completions \
  -H "Authorization: Bearer $OMNIAKEY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-6.1-sol",
    "messages": [
      {"role": "user", "content": "Review this pull request"}
    ],
    "stream": true
  }'

06

Estimate token cost

Enter expected input and output tokens to compare current listed rates.

OmniaKey estimate
$0.051786
Official-rate estimate
$0.40
Estimated savings
$0.348214

Estimate covers input and output tokens only. Cache usage, retries, tools, and client-side token accounting can change the final charge.

08

Sources and verification

Model facts and links were checked on 2026-10-27. Provider documentation remains the authority for upstream behavior.

Frequently asked questions

How much does GPT-6.1 Sol cost on OmniaKey?

Rates vary with input context length: <=272K: Input $0.225 USD, Output $1.464 USD, Cache hit / 1M $0.0093 USD; >272K: Input $0.45 USD, Output $2.929 USD, Cache hit / 1M $0.0186 USD.

What is the context window for GPT-6.1 Sol?

The listed context window is 1M. Actual usable limits can also depend on the client and upstream request format.

Is GPT-6.1 Sol better than Claude Opus 5.5 for coding?

The two vendors publish different evaluations on different harnesses, so neither set of results settles the question. Sol fits agents already on the OpenAI Responses API and gives an explicit effort setting from low to max. Opus 5.5 is worth a trial when diagnosis, rather than the edit, is the hard part. Run the same tasks and the same tests on both before picking a default.

How is GPT-6.1 Sol billed?

Per input and output token, with a separate cached-input rate for repeated prefixes; reasoning tokens count as output. Batch and Flex processing are listed below Standard and Fast mode above it. Trimming unused files and keeping the prefix stable both change the charge. Current rates appear in the price table on the model page.

Which endpoint and model id does GPT-6.1 Sol use?

Send POST /v1/chat/completions to https://api.omniakey.com/v1 with model set to gpt-6.1-sol. Tool calls need the Responses API at POST /v1/responses on the same base URL. The id is case-sensitive, so copy it from the model page. Codex, Cursor, Cline and Aider are listed as supported clients and take the base URL and key.

Can GPT-6.1 Sol read images?

Yes. The catalog lists image input alongside text, and output is text. A screenshot of a failing test, a chart from a report or a UI bug can travel in the same request as the code. Native audio and video input are not supported, so speech and screen recordings need a separate transcription step.

What changed from GPT-6 Sol to GPT-6.1 Sol?

Standard input and output rates are unchanged, while cache reads are listed lower and the release date moves to September 29, 2026. Tool calling now requires the Responses API, and the none and minimal effort levels are gone. A stronger answer can still consume more tokens, so a finished task may cost more.