LFM2.5-8B-A1B-Q4_K_M.gguf from LiquidAI/LFM2.5-8B-A1B-GGUF (credit to the original model authors and quantizer), split into <2 GB shards with llama-gguf-split so it can be loaded by wllama in the browser.

Load the first shard; wllama resolves the rest automatically.