GPT-6.1 Sol API pricing and cost calculator

By TechCompare · Updated · How we check

Sol's base rates make a 100K-input, 5K-output request $0.25 without cache hits or $0.155 with half the input cached. Check the long-prompt tier before scaling that estimate.

A newer version of GPT-6.1 Sol is out: GPT-6.2 Sol. Price GPT-6.2 Sol in the calculator or compare every OpenAI model.

Calculator

Cost Comparison

Based on 100,000 input tokens (50% cached), 5,000 output tokens, and 100 requests.Prices come live from OpenRouter at each maker's own price where it's listed. Temporary deals are left out.

Loading model prices

GPT-6.1 Sol's standard base rates are $2 per million input tokens and $10 per million output tokens. Cache reads cost $0.10 per million, so repeated prompt prefixes can make a meaningful difference to a session's input bill. The larger budgeting question is how many calls, output tokens, and retries a completed task needs. This page works through those token costs and the prompt-length boundary that changes the rates.

OpenAI: GPT-6.1 Sol

OpenAI's own price

Input
$2
Output
$10
Cached
$0.10

per 1M tokens

Batch availableFlex available

Price checked 2026-11-19

How this is calculated

For 100,000 input and 5,000 output tokens, an uncached base-tier request costs $0.20 for input plus $0.05 for output, or $0.25. The calculator's starting example reads half the input from cache, which puts the input estimate at $0.105 and the total at $0.155.

At 100 requests per month, that's $15.50 before cache writes and tools. Above 272,000 input tokens, the entire request moves to $4 input, $0.20 cache reads, and $15 output per million tokens. A 300,000-input, 5,000-output request then costs $1.275 without cache hits or $0.705 with half the input cached. These are standard provider text-token examples. The embedded calculator loads OpenRouter prices and its conditional tiers, so provider routing can affect the displayed estimate.

Verdict

Budget from completed tasks rather than one representative call. Record prompt length, cache-hit tokens, output usage, and attempts on your own workload, then apply the matching tier. Sol can be a useful candidate for coding and agent workflows that meet your quality targets at these rates. Keep unnecessary context out of each request, and compare larger document prompts separately because crossing the input boundary reprices the whole request.

More API Standalones scenarios

Frequently asked questions

How much would 10,000 GPT-6.1 Sol requests cost?
For 100,000 input and 5,000 output tokens per request, the standard token estimate is $2,500 with no cache hits or $1,550 with half the input read from cache. That example assumes identical usage on every call. Add cache-write charges, tool use, retries, and any processing or routing premiums when building a production budget.
What does the cache-read discount apply to?
It applies to input tokens actually read from cache, reducing their base rate from $2 to $0.10 per million, 95% off. It doesn't discount uncached input or generated output. Reading half the input from cache therefore doesn't cut the whole request price by 95%. The example request goes from $0.25 to $0.155. Cache writes have their own $2.50-per-million base rate, which the calculator doesn't include in its estimate of cache reads.
Does a large prompt change Sol's rates for the whole request?
Yes. When total input exceeds 272,000 tokens, the full request moves to $4 input, $0.20 cache reads, and $15 output per million, not just the tokens beyond the boundary. Below it, the rates are $2, $0.10, and $10. A prompt of exactly 272,000 input tokens stays in the base tier. Input that can be read from cache still contributes to prompt length, so caching doesn't remove the pricing boundary.
What does batch processing save on Sol?
OpenAI's Batch API prices applicable tokens 50% below the standard rate. The uncached 100K-input, 5K-output example becomes $0.125, while the half-cached token example becomes $0.0775 before write costs and tools. This is provider Batch API pricing. OpenRouter's interactive model list isn't a promise that the same request can be sent through OpenRouter as a batch job.
Is GPT-6.1 Sol pricing a ChatGPT subscription price?
No. These are usage-based API token rates, not a monthly ChatGPT plan fee. They describe model input, output, and caching categories for API calls. A budget also needs the usage generated by tools and reasoning, plus the number of attempts per completed task. Use provider billing records to reconcile the estimate with real spending.