Cerebras API pricing

Hosted7 language models

As of Oct 14, 2026, Cerebras lists 7 language models through its API. Its cheapest language model is Llama3.1 8B at $0.10 input / $0.10 output per million tokens.

Language models (7)

Prices per million tokens.

Model	Input	Output	Cached input	Context
gpt-oss-120b cerebras/gpt-oss-120b	$0.35	$0.75	—	131K
Llama 3.3 70B Instruct cerebras/llama-3.3-70b	$0.85	$1.20	—	128K
Llama3.1 8B cerebras/llama3.1-8b	$0.10	$0.10	—	128K
Llama3.1 70B cerebras/llama3.1-70b	$0.60	$0.60	—	128K
Qwen 3 32B cerebras/qwen-3-32b	$0.40	$0.80	—	128K
Qwen 3.8 27B cerebras/qwen-3.8-27b	$0.99	$1.49	—	66K
Gemma 4 31B cerebras/gemma-4-31b	$0.99	$1.49	—	131K

gpt-oss-120b
cerebras/gpt-oss-120b
Input
$0.35
Output
$0.75
Context
131K
Llama 3.3 70B Instruct
cerebras/llama-3.3-70b
Input
$0.85
Output
$1.20
Context
128K
Llama3.1 8B
cerebras/llama3.1-8b
Input
$0.10
Output
$0.10
Context
128K
Llama3.1 70B
cerebras/llama3.1-70b
Input
$0.60
Output
$0.60
Context
128K
Qwen 3 32B
cerebras/qwen-3-32b
Input
$0.40
Output
$0.80
Context
128K
Qwen 3.8 27B
cerebras/qwen-3.8-27b
Input
$0.99
Output
$1.49
Context
66K
Gemma 4 31B
cerebras/gemma-4-31b
Input
$0.99
Output
$1.49
Context
131K

Prices in USD from LiteLLM and models.dev, checked daily; confirm on Cerebras's pricing page before you rely on a number.