Using gpt-oss-20b on Replicate API

Implementation guide · GPT-OSS · OpenAI

ServerlessOpen Source

Replicate API exposes gpt-oss-20b through model ID openai/gpt-oss-20b. Use the setup steps, sourced pricing, capabilities, and official provider links below to validate this route before deployment.

Last refreshed 2026-04-19. Next refresh: weekly.

Quick Start

1
Create an account at Replicate API and generate an API key.
2
Use the Replicate API SDK or REST API to call openai/gpt-oss-20b — see the documentation for request format.
3
You'll be billed $0.09/1M input, $0.36/1M output tokens. See full pricing.

API Portal Documentation Pricing Model Card

Code Examples

Install

pip install replicate

API key

REPLICATE_API_TOKEN

Model ID

openai/gpt-oss-20b

Replicate uses "owner/model-name" format (e.g. "meta/meta-llama-3-8b-instruct") for the latest version, or "owner/model-name:version-sha" to pin to a specific version. The REST endpoint splits owner and model-name into the path: /v1/models/{owner}/{model-name}/predictions.

import replicate

# reads REPLICATE_API_TOKEN from env
# openai/gpt-oss-20b format: "owner/model-name" (latest version) or "owner/model-name:version-hash"
output = replicate.run(
    "openai/gpt-oss-20b",
    input={"prompt": "Hello"}
)
# Output is a list or generator depending on the model
print("".join(output))

Pricing on Replicate API

Type	Price (per 1M)
Input tokens	$0.09
Output tokens	$0.36

Capabilities

JSON / Tool useStructured Outputs

About gpt-oss-20b

OpenAI open-weight model with 20 billion parameters. Lightweight, efficient text-only model with reasoning and function calling. Free for self-hosting. Released August 5, 2025.

Full model details →

Model Specs

Released2025-08-05

Parameters20B

Context131k

ArchitectureDecoder Only

Knowledge cutoff2025-08

Replicate

San Francisco, California, United States