GPT OSS 20B pricing
$ Affordable$0.03/M input · $0.13/M output. List prices from provider data; in idapt, chat is metered per token with transparent, usage-based pricing at the prices shown in the app.
Full rate table
| Rate | Price | Unit |
|---|---|---|
| Input | 0.03 | /M tokens |
| Output | 0.13 | /M tokens |
| Cache read | 0.03 | /M tokens |
What typical work costs
| Workload | Tokens | List price |
|---|---|---|
| Quick question | 1K tokens in, 500 out | $0.000095 |
| Long-document summary | 60K tokens in, 2K out | $0.0021 |
| Agent working session | 400K tokens in, 40K out across a day | $0.02 |
Price your own workload
- Per call
- $0.000095
- Per month
- $0.0095
Compare up to four models at once in the cost calculator.
Frequently asked
What does GPT OSS 20B cost per 1M tokens?
GPT OSS 20B lists at $0.03 per 1M input tokens and $0.13 per 1M output tokens.
Does GPT OSS 20B discount cached input tokens?
Yes. Cached input tokens are billed at $0.03 per 1M, versus $0.03 for fresh input. Prompts that repeat a long prefix (a system prompt, attached files) benefit the most.
What does a typical chat message with GPT OSS 20B cost?
A message with about 1,000 input tokens and a 500-token reply costs roughly $0.0001 at list price. Long documents and long replies scale that linearly with token counts.
Is GPT OSS 20B cheaper than GPT Mini (120b)?
GPT OSS 20B has the lower blended price at a 3:1 input:output mix. GPT OSS 20B: $0.03 in / $0.13 out per 1M. GPT Mini (120b): $0.04 in / $0.17 out per 1M.
How does idapt bill GPT OSS 20B?
GPT OSS 20B is in idapt's free lane: signed-in users run it against the free daily allowance, with transparent, usage-based pricing beyond it at the prices shown in the app.
Part of the GPT OSS 20B model page · run it locally · cheapest models board · price changes