GPT-6.1 Sol API Cost Calculator
Calculate GPT-6.1 Sol API costs with long-context pricing, cached reads, cache writes, Fast, Flex, Batch, Ultrafast and EU regional processing.
Official OpenAI rates · Last verified October 10, 2026
GPT-6.1 Sol pricing
| Model | Input / 1M | Cached read / 1M | Cache write / 1M | Output / 1M |
|---|---|---|---|---|
| GPT-6.1 Sol | $2.00 | $0.10 | $2.50 | $10.00 |
Long context
| Model | Input / 1M | Cached read / 1M | Cache write / 1M | Output / 1M |
|---|---|---|---|---|
| GPT-6.1 Sol | $4.00 | $0.20 | $5.00 | $15.00 |
Worked workload examples
For 100K input and 10K output with one Standard short-context request, no cache: 100,000 × $2 / 1M + 10,000 × $10 / 1M = $0.30.
With 50% cached reads: 50,000 × $2 / 1M + 50,000 × $0.10 / 1M + 10,000 × $10 / 1M = $0.205.
With 50% reads and 10% writes: 40,000 × $2 / 1M + 50,000 × $0.10 / 1M + 10,000 × $2.50 / 1M + 10,000 × $10 / 1M = $0.210. The three input buckets sum to 100K.
300K input, 10K output, no cache: 300,000 × $4 / 1M + 10,000 × $15 / 1M = $1.35. Fast: $2.70; Flex or Batch: $0.675; EU Standard: $1.485; Ultrafast: $8.10.
Where the monthly bill actually comes from
Output volume first. At $10.00 per 1M, output costs five times fresh input, so verbosity is billed. A coding agent using 30,000 input and 8,000 output tokens per request with 40% cached reads costs $0.1172 per request — $2,344 a month at 20,000 requests — and $0.08 of each request is output alone.
The 272K cliff. Crossing 272,000 input tokens reprices the whole request at long-context rates, not just the tokens above the threshold. The worked example above prices 300K input at $1.35; the same request trimmed to 200K input would cost $0.50 at short rates. Prompt trimming near the threshold is worth real money.
Cache-write payback. A cache write costs $2.50 per 1M — 25× a cached read on this model. Against re-sending the prefix fresh at $2.00, a written prefix pays for itself on the first reuse; prefixes used only once are cheaper sent fresh. So: measure output length first, keep prompts under the long-context threshold where you can, and cache only prefixes you know will be re-read.
Current and previous Sol
GPT-6.1 Sol replaces GPT-6 Sol in the site's current lineup. The API model id is gpt-6.1-sol, with a 1,050,000-token context window and 128,000 maximum output tokens.
Compare GPT-6.1 Sol vs GPT-6 Sol → · Previous GPT-6 Sol calculator · What changed at release?
Frequently asked questions
How much does GPT-6.1 Sol cost?
Standard short-context rates per 1M tokens are $2.00 fresh input, $0.10 cached read, $2.50 cache write and $10.00 output.
When does long-context pricing apply?
Input above 272,000 tokens uses long rates for the full request: $4.00 fresh input, $0.20 cached read, $5.00 cache write and $15.00 output. Exactly 272,000 input tokens remains short context.
Does Fast mode work with EU regional processing?
Yes. Fast costs 2× Standard, and EU regional processing adds 10% on top; OpenAI lists EU data residency for GPT-6.1 Sol including Fast and Ultrafast modes. Ultrafast (6× Standard, added October 2026) also works with EU residency on this model — the calculator prices both combinations.
Is GPT-6.1 Sol always 50% cheaper than GPT-6 Sol?
No. Cached reads cost 50% less. Fresh input, cache writes and output have the same published rates. Total savings depend on the cached-read share and workload.
What does GPT-6.1 Sol cost per month for a coding agent?
For an illustrative coding agent — 30,000 input and 8,000 output tokens per request, 40% cached reads, 20,000 requests a month on Standard — about $2,344 a month. Output is $0.08 of the $0.1172 per-request cost, so response length is the biggest lever. Enter your own token counts in the calculator for your workload.
Official sources
GPT-6.1 Sol model documentation · OpenAI API pricing · Pricing methodology