Skip to content

d3:flash

6 TagsUpdated 9B params8192 contextEnglishApache-2.0by vLLM Semantic Router

Qwen3.5-9B: 0.712 on typed decisions, no more accurate than d3:mini at twice the memory.

9b
ollaya run d3:flash --preset triage "I was charged twice for my subscription this month and want a refund."

Details

  • graph45c7f6fc9ca9 · 12 MBonnx · d3 · 9B · fp32
  • weights2c322016b91b · 4.9 GBhuggingface.co/vllm-sr/d3-flash/resolve/13dc1cc…/model-00001-of-00004.safetensors
  • weightsf769650fb747 · 5.0 GBhuggingface.co/vllm-sr/d3-flash/resolve/13dc1cc…/model-00002-of-00004.safetensors
  • weightsc85a0f2f8df2 · 5.0 GBhuggingface.co/vllm-sr/d3-flash/resolve/13dc1cc…/model-00003-of-00004.safetensors
  • weightsd686ef3f6cde · 1.9 GBhuggingface.co/vllm-sr/d3-flash/resolve/13dc1cc…/model-00004-of-00004.safetensors
  • weights8bbc9a9df4e1 · 4 MBhuggingface.co/vllm-sr/d3-flash/resolve/13dc1cc…/readout.safetensors
  • tokenizer06b9509352d2 · 20 MBhuggingface.co/bespokelabs/Bespoke-Nimble-9B-v2/resolve/4b8c04d…/tokenizer.json
  • decision2411d4e98669 · 5 KB{"engine": "onnx", "family": "d3", "encoder": "", "layout": "d3-code-readout-v1", …}
  • calibration0f89e51fde90 · 133 B{"temperature": [1.0, 1.0, 1.0]}
  • license3d7ec11263ac · 10 KBDecision 3.0 (d3) by the vLLM Semantic Router contributors (https://huggingface.co/vllm-sr)

Every layer is checked against its sha256 when it is pulled. Weights and tokenizers download from the model author's Hugging Face repository at a pinned commit; Ollaya never re-hosts them.

Decision 3.0 (d3) readme and all models