Vision-preserving MLX conversion of deepreinforce-ai/Ornith-1.0-35B,
pinned to revision 5df2ed3f675c7beaa490328cc70bb573b65fb660.
This self-contained experimental variant combines the qualified MXFP4 Ornith body and
BF16 vision tower with a compatible Qwen3.5-MoE Q5/G64 MTP sidecar. Ornith's upstream
35B checkpoint does not include native MTP tensors. The sidecar is byte-identical to
mtp.safetensors from Shiftedx/ornith-1.0-35b-abliterated-mxfp4-vision-mtplx
at revision afed7f14a357c9db38bf67eb5b7c66b6152157f5 (SHA-256 d9788ed667167b211cdb178247f4d5ccc2213f349611d7c05486d23f6b364b84).
MTPLX 2.0.2 inspection and tensor-contract gates passed. In the frozen one-prompt performance test, depth 2 achieved 136.94 decode tok/s versus 89.41 tok/s autoregressive (+53.17%); end-to-end throughput improved from 82.89 to 123.84 tok/s (+49.40%). Depth-2 acceptance was 94.51% and 87.78% by draft depth. Quality gates passed, but every completion reached the 256-token ceiling, fans remained on automatic, and the figures are a point estimate rather than a statistical benchmark.
mtplx inspect --require-mtp <local-path-or-repo-id>
mtplx serve --model <local-path-or-repo-id> --generation-mode mtp --mtp-depth 2 --reasoning-mode off
python -m mlx_vlm.generate \
--model <local-path-or-repo-id> \
--image image.jpg \
--prompt "Describe this image." \
--max-tokens 256
Review the upstream model card for intended use, training, license, benchmarks, and deployment considerations. Quantization does not change the upstream model's safety characteristics.
model-00015-of-00016.safetensors 1.3 GBmodel-00012-of-00016.safetensors 1.3 GBmodel-00006-of-00016.safetensors 1.3 GBmodel-00009-of-00016.safetensors 1.3 GBmodel-00003-of-00016.safetensors 1.3 GBmodel-00008-of-00016.safetensors 1.2 GBmodel-00011-of-00016.safetensors 1.2 GBmodel-00005-of-00016.safetensors 1.2 GBmodel-00014-of-00016.safetensors 1.2 GBmodel-00002-of-00016.safetensors 1.2 GBmodel-00001-of-00016.safetensors 1.1 GBmodel-00013-of-00016.safetensors 1.1 GBmodel-00007-of-00016.safetensors 1.1 GBmodel-00010-of-00016.safetensors 1.1 GBmodel-00004-of-00016.safetensors 1.1 GBmodel-vision-00001-of-00001.safetensors 893 MBmodel-00016-of-00016.safetensors 445 MBornith-ai/Ornith-1.0-35B