Cudallm 8B
ByteDance's Cudallm 8B is a chat-based language model specifically designed for generating high-performance and syntactically correct CUDA kernels, exceling at assisting developers in writing and optimizing CUDA code for scientific computing and machine learning workloads. With a context window of 40,960 tokens, it has undergone a two-stage training process to master parallel programming complexities for GPUs.
Input
Output
Context
41K
Max Output
-
Parameters
8.2B
Technical Specifications
Model TypeChat
Context Window40,960 tokens
Max Output TokensNot available
Parameters8.2B
Release DateAug 3, 2025
Training CutoffNot available
Licenseapache-2.0
Open SourceYes
Input Modalities
Text
Output Modalities
Text
Capabilities
Estimated GPU Requirements for Cudallm 8B
Estimates based on INT8 quantization. Actual requirements vary by framework and configuration.
Related Tools
Data sourced from official provider APIs and documentation
Last updated: Jun 23, 2026
Start building with the right model.
Automatically route workloads to the right model for every task, every time.