How much does gpt-oss-20b cost via API?

It's free: $0 per 1M input tokens and $0 per 1M output tokens, both from the provider and on OpenKey. OpenKey applies a flat 3% fee on provider list price, but 3% of $0 is still $0, so there's no cost difference between using the provider directly or through OpenKey.

Is gpt-oss-20b free to use?

Yes, gpt-oss-20b is one of 25 free models on OpenKey, with $0 pricing for both prompt and completion tokens. It's also open-weight under Apache 2.0, so you can run it yourself via its Hugging Face repo (openai/gpt-oss-20b) instead of calling an API at all.

What is gpt-oss-20b's context window?

gpt-oss-20b has a 131,072 token context window, with a max completion length of 32,768 tokens per response. That context window is larger than 23% of models in OpenKey's catalog, enough room for long documents or extended multi-turn conversations in a single request.

Is gpt-oss-20b open source?

Yes. OpenAI released gpt-oss-20b as an open-weight model under the Apache 2.0 license, with weights published on Hugging Face at openai/gpt-oss-20b. It's a 21B parameter Mixture-of-Experts model with 3.6B active parameters per forward pass.

Does gpt-oss-20b support tool calling and structured outputs?

Yes, gpt-oss-20b supports tool calling (tools, tool_choice), structured outputs, and response_format parameters, along with reasoning controls, seed, and standard sampling options like temperature and top_p. Reasoning is mandatory on this model, with a default effort of medium and options for low or high.

gpt-oss-20b (free)

openaigpt-oss-20b:free

Free

gpt-oss-20b is an open-weight 21B parameter model released by OpenAI under the Apache 2.0 license. It uses a Mixture-of-Experts (MoE) architecture with 3.6B active parameters per forward pass, optimized for...

Try in playground Hugging Face ↗

Modalities: text → text
In / out per 1M: Free / Free
Context: 131K tokens
Added: Aug 5, 2025

Max output: 33K tokens
Tokenizer: GPT
Knowledge cutoff: Jun 30, 2024
Reasoning: always on

Pricing

Per 1M tokens. The provider price and our flat 3% fee are separate columns — what you pay is their sum.

Per 1M tokens	Provider	+ 3% fee	You pay
Input	Free	—	Free
Output	Free	—	Free

Providers

One company serves gpt-oss-20b (free). Figures are each provider’s own list price per 1M tokens — your OpenKey price is in the card above.

Provider	Context	Input /M	Output /M
Darkbloom	131K	Free	Free

Provider list from the OpenRouter catalog, Jul 28, 2026. All providers

Benchmarks

Artificial Analysis

Agentic index: 3.1
Coding index: 20.7
Intelligence index: 14.9

Design Arena

Category	Elo	Win rate	Rank
Data viz models	978	39.7%	#92
Websites models	895	27.9%	#107

Head-to-head preference voting. How we filter and rank

Supported parameters

frequency_penalty
include_reasoning
max_tokens
min_p
presence_penalty
reasoning
repetition_penalty
response_format
seed
stop
structured_outputs
temperature
tool_choice
tools
top_a
top_k
top_p

Call it

OpenAI-compatible: point your SDK at api.openkey.ai/v1 and use model openai/gpt-oss-20b:free.

curl Python TypeScript

curl https://api.openkey.ai/v1/chat/completions \
  -H "Authorization: Bearer ***" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt-oss-20b:free",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://api.openkey.ai/v1",
    api_key=os.environ["OPENKEY_API_KEY"],
)

completion = client.chat.completions.create(
    model="openai/gpt-oss-20b:free",
    messages=[{"role": "user", "content": "Hello"}],
)
print(completion.choices[0].message.content)

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.openkey.ai/v1",
  apiKey: process.env.OPENKEY_API_KEY,
});

const completion = await client.chat.completions.create({
  model: "openai/gpt-oss-20b:free",
  messages: [{ role: "user", content: "Hello" }],
});
console.log(completion.choices[0].message.content);

About gpt-oss-20b (free)

gpt-oss-20b is OpenAI's open-weight 21B parameter model, built with a Mixture-of-Experts architecture that activates only 3.6B parameters per forward pass, released under Apache 2.0 in August 2025. It runs with reasoning always on, at low, medium, or high effort, and takes text in, text out with a 131,072 token context window. Artificial Analysis puts its intelligence index at 14.9 and coding index at 20.7, positioning it as a lightweight, self-hostable option rather than a frontier-benchmark model. Knowledge cutoff is June 30, 2024.

This is a free model on OpenKey — $0 input, $0 output, so the 3% fee has nothing to apply to. It's one of OpenAI's few open-weight releases (Apache 2.0, Hugging Face id openai/gpt-oss-20b), against a lineup where GPT-4 runs $30/$60 per 1M tokens. Its 131,072 token context window beats 23% of catalog models. Reasoning is mandatory, not optional, with three selectable effort levels (low/medium/high), and it supports structured outputs and tool calling despite the small 3.6B active-parameter footprint.

Questions

How much does gpt-oss-20b cost via API?: It's free: $0 per 1M input tokens and $0 per 1M output tokens, both from the provider and on OpenKey. OpenKey applies a flat 3% fee on provider list price, but 3% of $0 is still $0, so there's no cost difference between using the provider directly or through OpenKey.
Is gpt-oss-20b free to use?: Yes, gpt-oss-20b is one of 25 free models on OpenKey, with $0 pricing for both prompt and completion tokens. It's also open-weight under Apache 2.0, so you can run it yourself via its Hugging Face repo (openai/gpt-oss-20b) instead of calling an API at all.
What is gpt-oss-20b's context window?: gpt-oss-20b has a 131,072 token context window, with a max completion length of 32,768 tokens per response. That context window is larger than 23% of models in OpenKey's catalog, enough room for long documents or extended multi-turn conversations in a single request.
Is gpt-oss-20b open source?: Yes. OpenAI released gpt-oss-20b as an open-weight model under the Apache 2.0 license, with weights published on Hugging Face at openai/gpt-oss-20b. It's a 21B parameter Mixture-of-Experts model with 3.6B active parameters per forward pass.
Does gpt-oss-20b support tool calling and structured outputs?: Yes, gpt-oss-20b supports tool calling (tools, tool_choice), structured outputs, and response_format parameters, along with reasoning controls, seed, and standard sampling options like temperature and top_p. Reasoning is mandatory on this model, with a default effort of medium and options for low or high.

More from OpenAI

All OpenAI models →

OpenAI

GPT Chat Latest

GPT Chat Latest points to OpenAI's stable API alias `chat-latest` that always resolves to the latest Instant chat model used in ChatGPT. As OpenAI rolls out new Instant model updates...

ctx 400K in $5.15 out $30.90

OpenAI

OpenAI GPT Latest

This model always redirects to the latest model in the OpenAI GPT family.

ctx 1.1M in $5.15 out $30.90

OpenAI

OpenAI GPT Mini Latest

This model always redirects to the latest model in the OpenAI GPT Mini family.

ctx 400K in $0.772 out $4.63

OpenAI

GPT-5.5

GPT-5.5 is OpenAI’s frontier model designed for complex professional workloads, building on GPT-5.4 with stronger reasoning, higher reliability, and improved token efficiency on hard tasks. It features a 1M+ token...

ctx 1.1M in $5.15 out $30.90