ornith-1.0-35b-mxfp8-mlx

Vision-preserving MLX conversion of deepreinforce-ai/Ornith-1.0-35B, pinned to revision 5df2ed3f675c7beaa490328cc70bb573b65fb660.

Format

• Language body: MXFP8, 8-bit, group size 32
• MoE gate overrides: affine 8-bit, group size 64
• Vision tower: source BF16, 333 tensors
• Architecture: Qwen3.5 MoE multimodal, 40 language layers, 256 experts
• Tokenizer, chat template, processor metadata, and vision features preserved from the pinned source

This standard artifact disables MTP metadata and does not include an MTP sidecar.

Local qualification

• Structural inspection: pass (1,658 indexed tensors; 512 quantized weights; 80 MoE gate overrides)
• Shard index: pass; no stale or missing shards
• Strict MLX-LM and MLX-VLM lazy load: pass
• Cross-variant tokenizer, template, processor, and vision-shard parity: pass
• Vision features: source BF16 tensor payload preserved

Basic use

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.