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

GPT-6.2 Sol API

NewLLM

The GPT-6.2 Sol API is a newer snapshot of OpenAI's cost-oriented flagship line, billed at 9% of OpenAI's list: 1M context, vision and tool use, with every meter following the list's step above 272K.

  • Input · <=272K$0.18 USDOfficial price$2 USD
  • Output · <=272K$0.9 USDOfficial price$10 USD
  • Input · >272K$0.36 USDOfficial price$4 USD
  • Output · >272K$1.35 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-6.2 Sol API is a newer snapshot of OpenAI's cost-oriented flagship line, billed at 9% of OpenAI's list: 1M context, vision and tool use, with every meter following the list's step above 272K.

GPT-6.2 Sol is a newer snapshot of the cheaper OpenAI line that the catalog carries as GPT-6.1 Sol, and it sits below GPT-6.2 Astra in the same family: same 1M window and the same vision and tool-use support, now on a cheaper card. The gateway bills this line at roughly a tenth of OpenAI's list price on every meter, and above 272K it follows the list's own step rather than a flat doubling. That makes it the cheapest Sol snapshot so far — well below the GPT-6.1 Sol entry on input, and further below it on output. The catalog quotes that step per band: one quote up to 272K and the list's step above.

In our 18 November 2026 runs a merge-intervals coding prompt returned 166, 168 and 147 output tokens over three calls, for a median of 166 — a little shorter than the same prompt on GPT-6.2 Astra in the same session. JSON mode returned a parseable object, tool calling produced a valid get_weather call for Lisbon, and image input correctly described a supplied cover on the first attempt. That mix is what the Sol line is for: predictable, cheap agent steps rather than the most expensive reasoning available.

This is the volume route. Use it for agent loops, batch jobs and everyday coding where Sol's rates already fit the budget and you want the newest snapshot instead of the previous one.

Best-fit coding workloads

High-volume agent loops

Use it for the many small steps inside an agent — classify, extract, decide, call a tool — and reserve the Astra snapshot for the steps that actually need reasoning depth. Practical tip: instrument cost per completed task, not cost per call, before assuming the cheaper line wins.

Batch extraction and tagging

Run thousands of documents through one fixed schema with JSON mode and keep going while the output parses. Practical tip: fail loudly on the first unparseable record instead of writing nulls, because a silent schema drift is far more expensive than a retry.

Everyday coding assistance

Wire it into Codex or Cursor for edits, test fixes and small refactors where the diff is short. Practical tip: keep the context to the file plus its direct callers; the window is large, but a small context makes the review smaller too.

02

Hands-on notes

Checked on 2026-11-18 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
166
Cost of that output at current prices
$0.00015
JSON mode
Passed
Tool calling
Passed
Image input
Passed
  • On 18 November 2026 three runs of a merge-intervals coding prompt returned 166, 168 and 147 output tokens, a median of 166.

  • The same session passed a JSON-mode request, produced a get_weather tool call for Lisbon, and described a supplied cover image correctly on the first attempt.

03

Technical specification

Catalog facts and integration details verified for this route.

Model ID
gpt-6.2-sol
Context
1M
Inputs and tools
Vision · Tool use
API endpoint
/v1/chat/completions
Cache hit / 1M
$0.009 USD
Coding clients
Codex · Cursor · Cline · Aider

Limits to plan around

  • The cheaper card is a price change, not a spec change: it answers like the previous Sol snapshot but costs less, so any cost comparison built on the old Sol rates needs re-running.
  • It is not the flagship: on the hardest reasoning tasks the Astra snapshot answered the same prompt with a little more elaboration, and we did not run a scored comparison.
  • Nothing is aliased. Clients must name the new id, and a setup that mixes Sol and Astra snapshots needs its own way to attribute spend per model.

04

Compare nearby models

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

ModelBest forContextInputOutput
GPT-6.2 SolThis page
gpt-6.2-sol
The newest Sol snapshot, billed at roughly a tenth of OpenAI's list price with the same long-context step; the default for volume work now.1M
$0.18 USD
Official price $2 USD
$0.9 USD
Official price $10 USD
GPT-6.1 Sol
gpt-6.1-sol
The previous Sol snapshot on the older, higher card; keep it as the baseline your evaluations compare against.1M
$0.225 USD
Official price $2 USD
$1.464 USD
Official price $10 USD
GPT-6.2 Astra
gpt-6.2-astra
The flagship sibling from the same launch: same window, higher rate card, and its own step above 272K. Route only the steps that need the extra reasoning here.1M
$0.72 USD
Official price $8 USD
$3.6 USD
Official price $40 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-6.2-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
$0.036
Official-rate estimate
$0.40
Estimated savings
$0.364

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

07

Related coding guides

Deeper selection, setup, and troubleshooting guidance for this model family.

08

Sources and verification

Model facts and links were checked on 2026-11-18. Provider documentation remains the authority for upstream behavior.

Frequently asked questions

How much does GPT-6.2 Sol cost on OmniaKey?

Rates vary with input context length: <=272K: Input $0.18 USD, Output $0.9 USD, Cache hit / 1M $0.009 USD; >272K: Input $0.36 USD, Output $1.35 USD, Cache hit / 1M $0.018 USD.

What is the context window for GPT-6.2 Sol?

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

How does GPT-6.2 Sol differ from GPT-6.2 Astra?

They share the launch date, the 1M window, vision and tool use, but not the rate card: Sol keeps the cheaper Sol rates and halved cached input, while Astra carries the flagship rates and a different discount in the catalog. Use Sol for volume and Astra for the steps where a wrong answer costs more than the tokens.

Does GPT-6.2 Sol cost more than GPT-6.1 Sol?

No — it is cheaper. GPT-6.2 Sol runs on its own lower card, billed at roughly a tenth of OpenAI's list price, while GPT-6.1 Sol keeps the older card, so input, output and cached input all cost less for the same kind of work; the price table on this page shows both cards per context band.

Should I switch everything to GPT-6.2 Sol?

Only after replaying the evaluation set. It answered our merge-intervals prompt slightly more tersely than GPT-6.2 Astra and passed JSON, tool calling and image input, but a snapshot can lose a case your tests cover and we cannot see. Keep a GPT-6.1 Sol baseline so a regression is visible immediately.

How was GPT-6.2 Sol checked before it was listed?

We ran a merge-intervals coding prompt three times on 18 November 2026 and recorded 166, 168 and 147 output tokens, then tested JSON mode, tool calling and image input on the same route. The counts are in the measurement table and the guide states what we did not test rather than rounding it away.