Freesolo-Co
/

Qwen3.6-27B-FP8

compressed-tensors

Model card Files Files and versions

Freesolo-Co/Qwen3.6-27B-FP8

Freesolo-owned FP8_DYNAMIC prequantized checkpoint for Qwen/Qwen3.6-27B. Generated with llmcompressor for vLLM-compatible inference. Vision tower + MoE router left in original precision (vision-language preserved).

Downloads last month: 259

Safetensors

Model size

27B params

Tensor type

BF16

·

F8_E4M3

·

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for Freesolo-Co/Qwen3.6-27B-FP8

Base model

Qwen/Qwen3.6-27B

Quantized

(715)

this model