GPT-OSS 20B (Groq) API

Pricing

Input / 1M	$0.07
Output / 1M	$0.30
Blended / 1M	$0.375
Cached input / 1M	$0.04
Context window	128,000 tokens
Max output	32,768 tokens
License	Open weights · Apache 2.0 permissive
Knowledge cutoff	June 2024 · source
Intelligence index	15.2 Artificial Analysis Intelligence Index · source

textfunction-calling

Price verified 2026-09-05 (0 days ago) · sourcecorroborated

Sources

Source	Input / 1M	Output / 1M	Verified
OpenRouter	$0.07	$0.30	2026-09-05
LiteLLM	$0.07	$0.30	2026-09-03
models.dev	$0.07	$0.30	2026-09-04

Two independent sources agree within 20%.

Latency measured

Region	TTFT p50	TTFT p95	Total p50	Total p95	Throughput	Samples
eu-frankfurt	390.3 ms	475.7 ms	532 ms	607 ms	1,869.7 tok/s	5
eu-paris	345.3 ms	419.2 ms	490.8 ms	529.8 ms	1,987 tok/s	5

Time-to-first-token, measured server-side. Measured 2026-09-04 (0 days ago).

Cheaper providers for gpt-oss-20b

Provider	Blended / 1M	Endpoint
AkashML	$0.12	view →
CoreWeave	$0.16	view →
DeepInfra	$0.17	view →

Compare all 12 providers serving gpt-oss-20b →

Call it

OpenAI-compatible endpoint — set GROQ_API_KEY.

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

from openai import OpenAI  # pip install openai

client = OpenAI(
    base_url="https://api.groq.com/openai/v1",
    api_key="...",  # GROQ_API_KEY
)
resp = client.chat.completions.create(
    model="openai/gpt-oss-20b",
    messages=[{"role": "user", "content": "Hello"}],
)
print(resp.choices[0].message.content)

Price history

Date	Input / 1M	Output / 1M
2026-09-01	$0.07	$0.30
2026-08-05	$0.07	$0.30

2 price points on record · a new point is added only when the price actually changes.

Use this data

Raw JSON for this page: gpt-oss-20b.json. Free to reuse under CC BY 4.0 with attribution to apipriceindex.com.