Quick answer
What does the gpt-oss-20b API cost when paid in crypto?
As of 23 October 2026, gpt-oss-20b costs $0.033 per million input tokens and $0.161 per million output tokens on LLMwithCrypto, under the API id gpt-oss-20b. That is 5% below the median of 2 hosts, with a 128K-token context. You top up in USDT, BTC or another coin and pay with no KYC.
How does our price compare with other hosts?
List prices per million tokens on each host's own pricing page, read on 23 October 2026; promotions are left out. Ours is the median of these prices minus 5%; every host is in the price index.
| Host | Input / 1M | Output / 1M |
|---|---|---|
| LLMwithCrypto | $0.033 | $0.161 |
| DeepInfra | $0.03 | $0.14 |
| Parasail | $0.04 | $0.20 |
Prices checked on
How do I call gpt-oss-20b?
With the OpenAI SDK, as for any OpenAI-compatible API: set the base URL to https://llmwithcrypto.com/api/v1, read your key from LLMWITHCRYPTO_API_KEY, and pass gpt-oss-20b as the model. Pass your API key as a Bearer token in the Authorization header, as the OpenAI SDK does.
import os
from openai import OpenAI
client = OpenAI(
base_url="https://llmwithcrypto.com/api/v1",
api_key=os.environ["LLMWITHCRYPTO_API_KEY"],
)
resp = client.chat.completions.create(
model="gpt-oss-20b",
messages=[{"role": "user", "content": "Extract the invoice number, date and total from this email as JSON."}],
)
print(resp.choices[0].message.content)
print(resp.usage) # prompt_tokens and completion_tokens: what the call costs
What is gpt-oss-20b good for?
gpt-oss-20b is the smaller of OpenAI’s two open-weight models, Apache 2.0, with a context of about 128K tokens. Only two of the hosts we read listed it on a public price page, so its median rests on fewer prices than most.
It fits small, frequent jobs where cost and speed matter most: extraction, routing, short answers and simple tool calls. For harder reasoning, gpt-oss-120b is the larger sibling on the same key.
What licence does gpt-oss-20b come with?
Apache 2.0 with OpenAI’s short usage policy (comply with applicable law), the same terms as gpt-oss-120b. Serving it as an API is allowed. Each model keeps its own licence and use policy (for example Meta’s Llama Acceptable Use Policy and the “Built with Llama” notice); using a model through us means accepting them.
gpt-oss-20b questions
Why only two hosts in the comparison?
On 23 October 2026, DeepInfra and Parasail were the only hosts we read that showed gpt-oss-20b on a public price page; the others did not list it, offered it for fine-tuning only, or had no readable price page.
Can I switch between gpt-oss-20b and 120b?
Yes, by changing the model field. Both run on the same endpoint and key, and each call is charged at the price of the model it used.
Is anything stored?
We do not store your prompts or the model’s answers: once a response is sent, its text is gone. Each call leaves one log line with the time, the account, the model and the token counts, deleted after 30 days.
Sources (3)
- DeepInfra pricing deepinfra.com/openai/gpt-oss-20b — DeepInfra (read 2026-10-23)
- Parasail pricing parasail.io/pricing — Parasail (read 2026-10-23)
- gpt-oss model card and licence — OpenAI on Hugging Face (read 2026-10-22)