# Unmatched Price Performance
Fast responses, scalable performance, and costs you can plan for.
## Large Language Models
| AI Model | Current Speed(Tokens per Second) | Input Token Price(Per Million Tokens) | Output Token Price(Per Million Tokens) |
| --- | --- | --- | --- |
| AI Model
GPT OSS 20B 128k | Current Speed
1,000 TPS | Input Token Price(Per Million Tokens)
$0.075(13.3M / $1)* | Output Token Price(Per Million Tokens)
$0.30(3.33M / $1)* |
| AI Model
GPT OSS Safeguard 20B | Current Speed
1,000 TPS | Input Token Price(Per Million Tokens)
$0.075(13.3M / $1)* | Output Token Price(Per Million Tokens)
$0.30(3.33M / $1)* |
| AI Model
GPT OSS 120B 128k | Current Speed
500 TPS | Input Token Price(Per Million Tokens)
$0.15(6.67M / $1)* | Output Token Price(Per Million Tokens)
$0.60(1.66M / $1)* |
| AI Model
Llama 4 Scout (17Bx16E) 128k | Current Speed
594 TPS | Input Token Price(Per Million Tokens)
$0.11(9.09M / $1)* | Output Token Price(Per Million Tokens)
$0.34(2.94M / $1)* |
| AI Model
Qwen3 32B 131k | Current Speed
662 TPS | Input Token Price(Per Million Tokens)
$0.29(3.44M / $1)* | Output Token Price(Per Million Tokens)
$0.59(1.69M / $1)* |
| AI Model
Llama 3.3 70B Versatile 128k | Current Speed
394 TPS | Input Token Price(Per Million Tokens)
$0.59(1.69M / $1)* | Output Token Price(Per Million Tokens)
$0.79(1.27M / $1)* |
| AI Model
Llama 3.1 8B Instant 128k | Current Speed
840 TPS | Input Token Price(Per Million Tokens)
$0.05(20M / $1)* | Output Token Price(Per Million Tokens)
$0.08(12.5M / $1)* |
*Approximate number of tokens per $
## Large Language Models (Enterprise-only)
| AI Model |
| --- |
| AI Model
Minimax M2.5 |
| AI Model
Qwen3-VL 32B |
## Text-to-Speech Models
| AI Model | Characters /s | Price (Per M Characters) |
| --- | --- | --- |
| AI Model
Canopy Labs Orpheus English | Characters /s
100 | Price
$22.00 |
| AI Model
Canopy Labs Orpheus Arabic Saudi | Characters /s
100 | Price
$40.00 |
## Automatic Speech Recognition (ASR) Models
| AI Model | Speed Factor | Price(Per Hour Transcribed) |
| --- | --- | --- |
| AI Model
Whisper V3 Large | Speed Factor
217x | Price
$0.111* |
| AI Model
Whisper Large v3 Turbo | Speed Factor
228x | Price
$0.04* |
*Audio is billed at a minimum of 10s per request.
## Prompt Caching
| Model | Uncached Input Tokens (Per M Tokens) | Cached Input Tokens (Per M Tokens) | Output Tokens (Per M Tokens) |
| --- | --- | --- | --- |
| Model
moonshotai/kimi-k2-instruct-0905 | Uncached Input Tokens (Per M Tokens)
$1.00 | Cached Input Tokens (Per M Tokens)
$0.50 | Output Tokens (Per M Tokens)
$3.00 |
| Model
openai/gpt-oss-120b | Uncached Input Tokens (Per M Tokens)
$0.15 | Cached Input Tokens (Per M Tokens)
$0.075 | Output Tokens (Per M Tokens)
$0.60 |
| Model
openai/gpt-oss-20b | Uncached Input Tokens (Per M Tokens)
$0.075 | Cached Input Tokens (Per M Tokens)
$0.0375 | Output Tokens (Per M Tokens)
$0.30 |
Note: No extra fee for the caching feature itself. The discount only applies when a cache hit occurs.
## Built-In Tools (Compound)
| Tool | Price | Parameter |
| --- | --- | --- |
| Tool
Basic Search | Price
$5 / 1000 requests | Parameter
web_search |
| Tool
Advanced Search | Price
$8 / 1000 requests | Parameter
web_search |
| Tool
Visit Website | Price
$1 / 1000 requests | Parameter
visit_website |
| Tool
Code Execution | Price
$0.18 / hour | Parameter
code_interpreter |
| Tool
Browser Automation | Price
$0.08 / hour | Parameter
browser_automation |
## Built-In Tools (GPT-OSS)
| Tool | Price | Parameter |
| --- | --- | --- |
| Tool
Browser Search - Basic Search | Price
$5 / 1000 requests | Parameter
browser_search - browser.search |
| Tool
Browser Search - Visit Website | Price
$1 / 1000 requests | Parameter
browser_search - browser.open |
| Tool
Code Execution - Python | Price
$0.18 / hour | Parameter
code_interpreter - python |
## About Our Pricing
No Surprise Inference Bills
Other inference providers spike costs without warning. Some hide behind elastic pricing. Groq pricing is linear and predictable, with no hidden costs or idle infrastructure. Every new user is growth, not risk, and you can keep margins secure.
## Compound Systems
Intelligent Tool Selection Across Multiple Models
Compound AI systems are powered by multiple openly-available models already supported in GroqCloud to intelligently and selectively use tools to answer user queries, starting first with web search and code execution. Pricing is passed through to the underlying models and server side tools that are part of the compound AI system.
## Batch API
Process Large-Scale Workloads Asynchronously
Batch processing lets you run thousands of API requests at scale by submitting your workload as an asynchronous batch of requests to Groq with 50% lower cost, no impact to your standard rate limits, and 24-hour to 7 day processing window.
## Build Fast
Seamlessly integrate Groq starting with just a few lines of code.