- Price
- gpt-oss-120b: $0.15 input / $0.60 output per million tokens
- Free plan
- Yes
- Setup
- Needs a developer
- Up and running
- Minutes for a free key; an afternoon for the first version in your app.
What Groq is for
Groq runs open AI models (like OpenAI's gpt-oss) on its own chips, built to answer very fast and cheap. It solves the "user gave up waiting" problem in chat bots and voice assistants. The trade-off is a short menu: you don't get Claude, GPT-6 or Gemini here.
How it plugs into your product
- 1Create an account on the Groq console and get a free API key, with per-minute and per-day limits.
- 2Build and test on a production model such as openai/gpt-oss-120b, calling it from your server.
- 3Upgrade to the Developer plan with a card: usage is billed afterwards, with early invoices at $1, $10, $100, $500 and $1,000 of usage.
- 4Set a spend limit and budget alerts, then check your real rate limits on the console's limits page.
Does it fit your idea?
It fits if
- A WhatsApp or voice bot where every second of delay loses the customer, and an open model handles the questions.
- High-volume simple tasks (classify tickets, extract fields) where $0.15 per million input tokens beats the big labs.
- You need cheap audio transcription: Whisper large v3 turbo is $0.04 per hour of audio.
Look elsewhere if
- Your product depends on top-tier reasoning or long documents: use Claude, GPT or Gemini.
- You plan to build on a preview model you saw in the catalog: Groq says previews can be dropped at short notice.
- You want to try several model families, open and closed, with one key: use OpenRouter.
What people only find out after building
Read this before you build on it. Each item links to the official page that says so.
The catalog is short. Production LLMs are Llama 3.1 8B, Llama 3.3 70B and gpt-oss 120B/20B, and the Llama ones say "contact sales" for price. Models like Qwen are preview only.
Official source: console.groq.comPreview models should not be used in production, because they may be discontinued at short notice.
Official source: console.groq.comBase limits are tight: the docs list gpt-oss-120b at 30 requests and 8,000 tokens per minute, and 200,000 tokens per day, per organization. A few simultaneous users can hit that; higher limits are for select workloads.
Official source: console.groq.comPayment is automatic from a credit card, US bank account or SEPA debit. If a payment fails and isn't fixed, the service may be suspended.
Official source: console.groq.com
Price
gpt-oss-120b: $0.15 input / $0.60 output per million tokens
Free plan with per-model rate limits per minute and per day.
- Price per million tokens (main model)
- gpt-oss-120b: $0.15 in / $0.60 out
- Models
- Open models: Llama, gpt-oss, Whisper
- Free tier
- Rate-limited free plan
Checked on Oct 28, 2026 · official pricing
What it costs in practice
Estimate: 1,000 users × 30 chats a month × about 2,000 tokens per chat (1,500 in, 500 out) = 45M input and 15M output tokens. On gpt-oss-120b ($0.15 in / $0.60 out per million): 45 × $0.15 + 15 × $0.60 ≈ $16/month. The bill is small, but the base limit of 200,000 tokens a day (about 6M a month) covers a tenth of this volume, so check your real limits first.
Estimate from the official price. Check it before you decide.
At a glance
- Free plan: yes
- Open source: no
“Not stated” means the official site doesn’t say.
Alternatives
Compare side by side
Faundr
Picked your tools? Build it with AI, safely.
Faundr keeps your project’s memory, blocks what can’t be undone and tells you if it’s safe to launch, inside Claude Code and Codex.