gpt-6-sol
Built to power complex coding and agentic workflows.
Modalities
In / Out Price
40% off$1.2/ 1M
$2 Official price
$6/ 1M
$10 Official price
Context
1.05M
Released
Sep 22, 2026
API call examples
POST /v1/chat/completionscurl --request POST 'https://api.tokatlas.com/v1/chat/completions' \
--header "Authorization: Bearer $UNIROUTER_API_KEY" \
--header 'Content-Type: application/json' \
--data '{
"model": "gpt-6-sol",
"messages": [
{
"role": "user",
"content": "Explain quantum entanglement."
}
],
"temperature": 0.7
}'Providers & routes
Current provider · OpenAI
Platform configured price · Group: codex
USD / 1M tokens. Each row shows input and output from the same tier. Displayed rates use the lowest available group multiplier; your API key group may differ.
GPT-6 Sol for connecting specs and code changes
Use it for code review, document analysis, and multi-step tool workflows. Give it a reproducible sample and clear completion criteria.
Long context
Review code, tests, and specs together to verify cross-file behavior.
Tunable reasoning
Evaluate routine coding and multi-turn reasoning separately.
Code and tools
Verify proposals against tool results and narrow the scope of changes.
Text and images
Ask for visual evidence before drawing conclusions about differences from the spec.
About
Cross-file coding and agent tasks
Sol accepts text and images and returns text. Roughly 1.05M context and 128K output leave room for modules, specs, screenshots, and tests. Use real tool results to verify and revise its proposals.
Sol, Luna, and Astra
Control test data, tools, and acceptance checks across candidates. Evaluate routine processing separately from complex cross-module work. Track handoffs, retries, and completion time for long tasks.
Tool compatibility before migration
If the workflow sends function tools, test supported effort-and-tool combinations on the route you use. Cover streaming, errors, and completion conditions. Validate tool arguments against their schema.
Start repository work with the smallest failing test. If tools depend on a no-reasoning response, exercise actual tool calls, requests without tools, and streaming on the endpoint. A short text-only smoke test will not cover those paths.
gpt-6-sol FAQ
Is Sol suitable for coding tasks that require repeated verification?
Start with one task backed by clear tests. Keep tool permissions, execution logs and completion criteria in the application.
How can I check whether reasoning works with function calling?
Confirm the model's route and test it with one simple tool. Parameter support can differ between Responses and Chat Completions.
How should I compare Sol with another model before migrating?
Run a set of tasks with identical prompts, data and acceptance tests. Include retries and fixes when measuring total time and cost.
Can Sol help when the deliverable is an image?
Use Sol to organize the brief or draft an image prompt, then send it to a dedicated image model. Validate the actual file before delivery.