Groq pricing
Groq built its own inference chip, the LPU, instead of buying GPUs, and sells that speed advantage as a per-token API for open-weight models. It is the cheapest of the three inference vendors tracked here on a per-token basis for small models.
- Priced by
- Unknown
- Headquarters
- Mountain View, US
- Founded
- 2016
- Pricing page tracked since
- Unknown
- Last checked
- Sep 21, 2026
Groq price history and price changes
Groq's pricing was read on Sep 2026. Prices are USD per million input tokens.
Llama 3.1 8B Instant has been listed since Sep 2026, starting at $0.05.
GPT OSS 120B has been listed since Sep 2026, starting at $0.15.
Llama 3.3 70B Versatile has been listed since Sep 2026, starting at $0.59.
First read on 21 September 2026, so the chart starts now.
Plans and prices
Prices as extracted on Sep 21, 2026.
Llama 3.1 8B Instant
$0.05per million input tokens, $0.08 per million output
Sep 21, 2026
GPT OSS 120B
$0.15per million input tokens, $0.60 per million output
Sep 21, 2026
Llama 3.3 70B Versatile
$0.59per million input tokens, $0.79 per million output
Sep 21, 2026
Compare Groq with its category
Alternatives to Groq
Other products people compare when shopping for this one.
Pricing page over the years
No archived copies of this pricing page were found yet.
See how other products lay out their plans in our pricing page examples.
Similar companies
Other tracked companies close to Groq by industry, go-to-market motion or location.
Groq pricing questions
- How much does Groq cost?
- Groq plans start at $0.05 per million input tokens, $0.08 per million output for Llama 3.1 8B Instant and go up to $0.59 per million input tokens, $0.79 per million output for Llama 3.3 70B Versatile. Prices are as shown on the pricing page when it was last read on Sep 21, 2026.
- Has Groq raised its prices?
- Llama 3.1 8B Instant has been listed since Sep 2026, starting at $0.05. GPT OSS 120B has been listed since Sep 2026, starting at $0.15. Llama 3.3 70B Versatile has been listed since Sep 2026, starting at $0.59. First read on 21 September 2026, so the chart starts now.
About Groq
Groq have offered 3 different plans starting with Llama 3.1 8B Instant at 0.05$. Groq is a Mountain View, US based company. Founded in 2016. They are using a combination of Sales-Led and Product-Led go-to-market strategy. The company describes themselves as "Groq runs open-weight models on its own LPU chips instead of GPUs, selling that speed as a per-token API. Rates scale with model size, from Llama 3.1 8B at a few cents per million tokens up to Llama 3.3 70B, with a free tier for prototyping.". They are in involved in Artificial Intelligence, Developer Tools, API, Cloud Computing, SaaS.