FluidInference / d3-nano-coreml

huggingface.co
Total runs: 62
24-hour runs: 2
7-day runs: 10
30-day runs: 10
Model's Last Updated: October 12 2026
zero-shot-classification

Introduction of d3-nano-coreml

Model Details of d3-nano-coreml

d3-nano — Core ML

vllm-sr/d3-nano (2.2B, part of Decision 3.0 by the vLLM Semantic Router team) converted to Core ML for Apple silicon: text, image and video requests on device. Same contract as upstream: one forward pass per question, the 255-way FP32 readout at the last prompt token, a probability for every option, no text generated.

Files
File What
d3_E256 / d3_E640 / d3_E1536.mlpackage text graph (fp16 compute, int8 weights) per prompt-length bucket; input = token embeddings with image/video tokens spliced in
vision_P320 / vision_P1536 / vision_P3072.mlpackage vision tower (fp32) per patch bucket; video frames go through it one frame pair at a time (block mask)
embeddings.f16 , pos_embed.f32 , readout.f32 , tokenizer.json , config.json , coreml_config.json host-side tables and config
python/ Python host (coremltools): d3_coreml_text.py , d3_coreml_image.py , d3_coreml_video.py

Swift: D3VisionManager in FluidUse loads this folder.

Checked against upstream

Same weights as upstream v3.1.0. Answers that differ from the upstream PyTorch runtime (bf16, MPS), same requests:

flips
Text (typed-decisions TEST, 20 requests) 0 / 100
Images (receipt checks) 0 / 24
Video (4-second street clips, 2 fps) 1 / 28

Speed on an M5 Pro (24 GB), CPU_AND_GPU : ~65 ms per text question, 0.57 s per 4-second video look.

Python
from d3_coreml_text import D3CoreMLText      # d3_coreml_image.D3CoreMLImage, d3_coreml_video.D3CoreMLVideo
m = D3CoreMLText("<vllm-sr/d3-nano snapshot>", "<this repo>")
m.system_one(state="...", questions={"route": {"type": "choice", "criteria": {...}}})

The Python host builds prompts with upstream's own runtime ( d3_runtime.py in the upstream snapshot, run without weights), so prompts match upstream exactly.

License: Apache-2.0, as upstream.

Runs of FluidInference d3-nano-coreml on huggingface.co

62
Total runs
2
24-hour runs
4
3-day runs
10
7-day runs
10
30-day runs

More Information About d3-nano-coreml huggingface.co Model

More d3-nano-coreml license Visit here:

https://choosealicense.com/licenses/apache-2.0

d3-nano-coreml huggingface.co

d3-nano-coreml huggingface.co is an AI model on huggingface.co that provides d3-nano-coreml's model effect (), which can be used instantly with this FluidInference d3-nano-coreml model. huggingface.co supports a free trial of the d3-nano-coreml model, and also provides paid use of the d3-nano-coreml. Support call d3-nano-coreml model through api, including Node.js, Python, http.

FluidInference d3-nano-coreml online free

d3-nano-coreml huggingface.co is an online trial and call api platform, which integrates d3-nano-coreml's modeling effects, including api services, and provides a free online trial of d3-nano-coreml, you can try d3-nano-coreml online for free by clicking the link below.

FluidInference d3-nano-coreml online free url in huggingface.co:

https://huggingface.co/FluidInference/d3-nano-coreml

d3-nano-coreml install

d3-nano-coreml is an open source model from GitHub that offers a free installation service, and any user can find d3-nano-coreml on GitHub to install. At the same time, huggingface.co provides the effect of d3-nano-coreml install, users can directly use d3-nano-coreml installed effect in huggingface.co for debugging and trial. It also supports api for free installation.

d3-nano-coreml install url in huggingface.co:

https://huggingface.co/FluidInference/d3-nano-coreml

Url of d3-nano-coreml

Provider of d3-nano-coreml huggingface.co

FluidInference
ORGANIZATIONS

Other API from FluidInference