Small model. Serious local speed.

LFM2.5 230M from Liquid AI, presented by BlackRiver Labs. High-speed language inference runs entirely in your browser through WebGPU. No prompt leaves your machine.

INITIALIZING LFM2.5

Fetching LFM2.5 weights0%

Checking WebGPU…

Ready · on-device

Fast thinking. Private by design.

BlackRiver local runtime · LFM2.5 runs entirely on your device

Kernels

What are Kernels?

Kernels are the low-level GPU programs that do the model's actual math — the matrix multiplications, attention, and normalization behind every token. And how well they're optimized can dramatically speed up inference.

WebGPU & WGSL. Each kernel is a WebGPU compute shader, written in WGSL — the language that runs general-purpose math on the GPU — entirely locally in your browser.
Agentic Kernel Optimization. Every kernel was generated by AI (in this case, Fable 5 and Opus 4.8), benchmarked on an Apple M4 Max, and refined through an evolutionary, genetic-style search toward the fastest version.
Blazingly Fast. This means we are able to run LFM2.5 230M at ~1400 tokens/sec on an M4 Max, pushing your device to its limits.

Select a kernel to read its real source