Does GPT OSS 20B support zero data retention?

Yes. GPT OSS 20B is available with zero data retention by default on AWS Bedrock, Fireworks, Geodd, Groq, Novita through the Opper gateway.

Where is GPT OSS 20B hosted?

GPT OSS 20B runs on both EU and US hosting routes through the Opper gateway. Pick the route that matches your residency requirement.

Is GPT OSS 20B trained on prompts and outputs?

No. Every hosting route for GPT OSS 20B is configured so customer prompts and outputs are not used to train models by default.

Is a GDPR DPA available for GPT OSS 20B?

Yes. A GDPR Data Processing Agreement is available on at least one route serving GPT OSS 20B. Contact hello@opper.ai to receive it.

GPT OSS 20B

by OpenAI

GPT-OSS 20B is an open-weight Mixture-of-Experts language model released by OpenAI in August 2025 under the Apache 2.0 license, alongside the larger 120B variant. It has 20.9B total parameters with 3.6B active per token across 24 layers, and can run on devices with around 16GB of memory. The model matches or exceeds OpenAI o3-mini on common benchmarks while staying highly efficient. It supports reasoning, tool use with function calling, and structured outputs, fitting cost-conscious agentic workflows and local inference. Its MoE design selects the top 4 of 32 experts per token, with a 128k context window, giving developers a compact open-weight option that keeps strong reasoning capability.

Open in playground Try in chat Compare

Key info

Input

Output

Features

Context window

128K

Max output

Input price

$0.07 /1M

Output price

$0.30 /1M

EU/EEA residency available
US residency available
Zero data retention on pay-as-you-go
No training by default
GDPR DPA available

Available routes

GPT OSS 20B runs on 6 different routes through the Opper gateway. Compare residency, ZDR, and training posture at a glance — full data-handling detail per route below.

Provider	Region	Zero data retention	Training	Input	Output
AWS Bedrock	EU	Zero data retention	No	$0.07	$0.30
Fireworks	US	Enterprise	No	$0.07	$0.30
Geodd	EU/EEA	Zero data retention	No	$0.03	$0.14
Geodd	US	Zero data retention	No	$0.03	$0.14
Groq	US	Enterprise	No	$0.07	$0.30
Novita	US	Enterprise	No	$0.04	$0.15

Training posture across routes: No training on prompts by default.

Data handling per route

Each route hosting GPT OSS 20B has its own privacy posture, residency, and GDPR terms. Postures are maintained by Opper with a last-verification timestamp.

AWS Bedrock — Sweden🇸🇪

Zero data retention is on by default on Pay-as-you-go — no action required. No training on customer data. EU; DPA available.

Zero data retention: On by default on Pay-as-you-go.
Training: No training on customer data.
Logging: None
Third-party access: None disclosed
GDPR DPA: DPA available
Transfer mechanism: Not applicable — data stays in EU

Fireworks — United States🇺🇸

Zero data retention is available via Opper Enterprise contract. No training on customer data. US; SCCs; DPA available.

Zero data retention: Available via Opper Enterprise contract.
Training: No training on customer data.
Logging: None
Third-party access: None disclosed
GDPR DPA: DPA available
Transfer mechanism: SCCs

Geodd — Norway🇳🇴

Zero data retention is on by default on Pay-as-you-go — no action required. No training on customer data. EEA; SCCs.

Zero data retention: On by default on Pay-as-you-go.
Training: No training on customer data.
Logging: None
Third-party access: Provider may share with subprocessors / partners
GDPR DPA: No DPA
Transfer mechanism: SCCs

Geodd — United States🇺🇸

Zero data retention is on by default on Pay-as-you-go — no action required. No training on customer data. US; SCCs.

Zero data retention: On by default on Pay-as-you-go.
Training: No training on customer data.
Logging: None
Third-party access: Provider may share with subprocessors / partners
GDPR DPA: No DPA
Transfer mechanism: SCCs

Groq — United States🇺🇸

Zero data retention is available via Opper Enterprise contract. No training on customer data. US; SCCs; DPA available.

Zero data retention: Available via Opper Enterprise contract.
Training: No training on customer data.
Logging: Abuse monitoring (30-day retention)
Third-party access: Provider may share with subprocessors / partners
GDPR DPA: DPA available
Transfer mechanism: SCCs

Novita — United States🇺🇸

Zero data retention is available via Opper Enterprise contract. No training on customer data. US; SCCs; DPA available.

Zero data retention: Available via Opper Enterprise contract.
Training: No training on customer data.
Logging: Abuse monitoring
Third-party access: None disclosed
GDPR DPA: DPA available
Transfer mechanism: SCCs

Benchmarks

Independent benchmark scores — composite indices for reasoning, coding, and math, plus individual eval scores where available.

Global rank#250 of 531 LLMs

TierEfficient

Output speed227 tok/s

First token0.44s

Intelligence Index14.9

Coding Index18.5

Math Index89.3

Reasoning & knowledge

MMLU-Pro

75%

GPQA Diamond

69%

Humanity's Last Exam

10%

Long-context reasoning

31%

Coding

LiveCodeBench

78%

SciCode

34%

Agentic & tool use

Terminal-Bench Hard

11%

τ²-Bench Telecom

60%

Math & instruction following

AIME 2025

89%

IFBench

65%

See full leaderboard →Benchmarks via Artificial Analysis · View on AA

Get started

Call GPT OSS 20B through the Opper gateway with one API key. Let your coding agent set it up, or call it directly — Opper is drop-in compatible with the OpenAI, Anthropic, and Google AI SDKs.

Set it up with your agent

Copy this and paste it into your coding agent — Claude Code, Cursor, Codex, and more — and it'll wire up Opper for you.

Or call it directly

1import OpenAI from "openai";
2
3const client = new OpenAI({
4  apiKey: process.env.OPPER_API_KEY,
5  baseURL: "https://api.opper.ai/v3/compat",
6});
7
8const completion = await client.chat.completions.create({
9  model: "aws/gpt-oss-20b-eu",
10  messages: [{ role: "user", content: "Hello" }],
11});
12console.log(completion.choices[0].message.content);