ClikaRT
ClikaRT is a C++ inference runtime with Python and Kotlin bindings that loads and serves AI models on CPUs, GPUs and other accelerators through one API.
What it does
- Model inference
Features 5
- Run-time backend selectionThe backend is picked at run time, not build time, across CPU, CUDA, Vulkan and Metal.
- PyTorch-shaped APIThe C++ API is PyTorch-shaped, with Python bindings for all of it and Kotlin for inference.
- Quantized weightsWeight-only quantization schemes such as GGUF block formats, FP8 and NF4 decode inside the kernels.
- Model loadingLoads safetensors, GGUF, ONNX and NumPy models through the ClikaRT io module.
- Serving runtimeProvides sessions, continuous batching and pipelines, plus an HTTP server and client.
Sources
- Pricing
-
No price could be checked on the pages read.
- Description
-
Official documentation · 10 Nov 2026
Product page: "ClikaRT is the CLIKA inference runtime: a C++ library, with Python and Kotlin bindings, that loads models and runs them on CPUs, GPUs and other hardware accelerators through one public API."
- What it does
-
Official documentation · 10 Nov 2026
Product page: "ClikaRT is the CLIKA inference runtime: a C++ library, with Python and Kotlin bindings, that loads models and runs them on CPUs, GPUs and other hardware accelerators through one public API."
- Sold within
-
No standalone plan could be checked on the pages read; sold through the vendor's sales channel.
- Name
-
Official documentation · 10 Nov 2026
Name as published on the product page.
- URL
-
Official documentation · 10 Nov 2026
Product page, read 2026-11-10.
- Deployment
-
Official documentation · 10 Nov 2026
Product page: "Run your models on phones, edge devices and cloud instances."
- Interfaces
-
Official documentation · 10 Nov 2026
Product page: "Run your models on phones, edge devices and cloud instances."
Something wrong here? Send a correction — quote this product id: clika-rt.