LLM ReferenceLLM Reference

Code Llama

AI at MetaCodingHighlight
12 models2023–2024Up to 100K ctxFrom $0.05/1M input

About

Code Llama is a family of large language models developed by Meta AI, designed for code generation and understanding tasks. Building upon the Llama 2 architecture, Code Llama offers variations to accommodate diverse needs, featuring model sizes with 7B, 13B, 34B, and 70B parameters that adjust complexity and capability. The family comprises the foundational model for general code tasks, Code Llama - Python for tasks specific to the Python language, and Code Llama - Instruct, which is fine-tuned for interpreting natural language instructions. Trained on extensive code and related data, these models excel in code completion, debugging, and generating code from natural language prompts. Larger models tend to provide enhanced performance, albeit at the cost of increased computational demands 12.

Specifications(12 models)

Code Llama model specifications comparison
ModelReleasedContextParametersStructured Outputs
CodeLlama 70B Python2024-0116K70BYes
CodeLlama 70B Instruct2024-0116K70BNo
CodeLlama 34B2023-08100K34BYes
CodeLlama 13B2023-08100K13BYes
CodeLlama 7B2023-08100K7BYes
CodeLlama 34B Python2023-08100K34BYes
CodeLlama 13B Python2023-08100K13BYes
CodeLlama 7B Python2023-08100K7BYes
CodeLlama 34B Instruct2023-0816K34BNo
CodeLlama 13B Instruct2023-0816K13BNo
CodeLlama 7B Instruct2023-0816K7BNo

Available From(8 providers)

Pricing

Code Llama model pricing by provider
ModelProviderInput / 1MOutput / 1MType
CodeLlama 7BReplicate API$0.05$0.25Serverless
CodeLlama 7B PythonReplicate API$0.05$0.25Serverless
CodeLlama 13BReplicate API$0.1$0.5Serverless
CodeLlama 13B PythonReplicate API$0.1$0.5Serverless
CodeLlama 7BTogether AI$0.2$0.2Serverless
CodeLlama 34BDeepInfra$0.2$0.45Serverless
CodeLlama 7B PythonTogether AI$0.2$0.2Serverless
CodeLlama 13BFireworks AI$0.2$0.2Provisioned
CodeLlama 13B PythonFireworks AI$0.2$0.2Provisioned
CodeLlama 7BFireworks AI$0.2$0.2Provisioned
CodeLlama 7B PythonFireworks AI$0.2$0.2Provisioned
CodeLlama 13B InstructFireworks AI$0.2$0.2Serverless
CodeLlama 7B InstructFireworks AI$0.2$0.2Serverless
CodeLlama 34BReplicate API$0.2$1Serverless
CodeLlama 34B PythonReplicate API$0.2$1Serverless
CodeLlama 13BTogether AI$0.3$0.3Serverless
CodeLlama 13B PythonTogether AI$0.3$0.3Serverless
CodeLlama 7BMicrosoft Foundry$0.52$0.67Provisioned
CodeLlama 7B PythonMicrosoft Foundry$0.52$0.67Provisioned
CodeLlama 70B PythonReplicate API$0.65$2.75Serverless
CodeLlama 34BTogether AI$0.8$0.8Serverless
CodeLlama 34B PythonTogether AI$0.8$0.8Serverless
CodeLlama 13BMicrosoft Foundry$0.81$0.94Provisioned
CodeLlama 13B PythonMicrosoft Foundry$0.81$0.94Provisioned
CodeLlama 70B PythonTogether AI$0.9$0.9Serverless
CodeLlama 34BFireworks AI$0.9$0.9Provisioned
CodeLlama 34B PythonFireworks AI$0.9$0.9Provisioned
CodeLlama 70B PythonFireworks AI$0.9$0.9Provisioned
CodeLlama 34B InstructFireworks AI$0.9$0.9Serverless
CodeLlama 70B InstructFireworks AI$0.9$0.9Serverless
CodeLlama 34BMicrosoft Foundry$1.54$1.77Provisioned
CodeLlama 34B PythonMicrosoft Foundry$1.54$1.77Provisioned
CodeLlama 34BIBM watsonx$1.8$1.8Serverless
CodeLlama 70B PythonMicrosoft Foundry$3.78$11.34Provisioned

Frequently Asked Questions

What is Code Llama?
Code Llama is a family of large language models developed by Meta AI, designed for code generation and understanding tasks. Building upon the Llama 2 architecture, Code Llama offers variations to accommodate diverse needs, featuring model sizes with 7B, 13B, 34B, and 70B parameters that adjust complexity and capability. The family comprises the foundational model for general code tasks, Code Llama - Python for tasks specific to the Python language, and Code Llama - Instruct, which is fine-tuned for interpreting natural language instructions. Trained on extensive code and related data, these models excel in code completion, debugging, and generating code from natural language prompts. Larger models tend to provide enhanced performance, albeit at the cost of increased computational demands 12.
How many models are in the Code Llama family?
The Code Llama family contains 12 models.
What is the latest Code Llama model?
The latest model is CodeLlama 70B, released in 2024-01.
How much does Code Llama cost?
Code Llama models range from $0.05/1M to $3.78/1M input tokens depending on the model and provider.

Models(12)