Llama 3.3 70B Instruct

Meta
llama-3.3-70b-instruct
Released Dec 6, 2024
Providers
4
Context
131K
Input
$0.10/M
Output
$0.32/M

Meta's powerful 70B parameter model quantized to FP8 for fast inference. Excellent for complex reasoning and multilingual tasks.

AWS BedrockDeprecating on AWS Bedrock

bedrock/llama-3.3-70b-instructQuantUnspecified
Input
$0.72/M
Output
$0.72/M
Context
128K
Max out
4K
Caching
Not offered
Thinking out
$0.72/M
FlexPriority / Fast
cloudflare/llama-3.3-70b-instructQuantUnspecified
Input
$0.29/M
Output
$2.25/M
Context
24K
Max out
6K
Caching
Not offered
Thinking out
$2.25/M
FlexPriority / Fast
deepinfra/llama-3.3-70b-instructQuantfp8
Input
$0.10/M
Output
$0.32/M
Context
131K
Max out
33K
Caching
Not offered
Thinking out
$0.32/M
Priority / Fast
novita/llama-3.3-70b-instructQuantbf16
Input
$0.14/M
Output
$0.40/M
Context
131K
Max out
12K
Caching
Not offered
Thinking out
$0.40/M
FlexPriority / Fast