Code Llama
A code-specialized large language model released by Meta. It follows a training strategy of initial pretraining on general-purpose corpora followed by fine-tuning on code-specific data, resulting in a model optimized for code generation, completion, and understanding tasks.
Core Facts
Timeline (last 90 days)
Meta's Code Llama and Mistral's Codestral are open-source code models that can be self-hosted.
Open-source models such as locally deployed Code Llama and DeepSeek Coder can be combined to handle non-core tasks, reducing reliance on paid quotas.
Developers can use tools such as Ollama and LM Studio to run open-source code models (such as DeepSeek Coder and Code Llama) locally as a backup solution; their capability ceiling is below that of top cloud-based models but they are fully controllable.
The latest version of Meta's Llama series is Llama 4.
All Facts (6)
Pass rates of code generation models on benchmarks such as HumanEval jumped from 60% to over 90% between 2024 and 2025
80%UnverifiedContinue是一个开源AI编程助手框架，允许开发者自由选择后端模型，包括本地部署的Code Llama、DeepSeek Coder等开源模型
85%UnverifiedDevelopers can use tools such as Ollama and LM Studio to run open-source code models (such as DeepSeek Coder and Code Llama) locally as a backup solution; their capability ceiling is below that of top cloud-based models but they are fully controllable.
60%UnverifiedThe latest version of Meta's Llama series is Llama 4.
60%UnverifiedMeta's Code Llama and Mistral's Codestral are open-source code models that can be self-hosted.
50%UnverifiedOpen-source models such as locally deployed Code Llama and DeepSeek Coder can be combined to handle non-core tasks, reducing reliance on paid quotas.
50%