Ornith 1.5 9B
9.7B parameter open-weight model. Dense 9.65B on the Qwen3.5 architecture, 256K context, image input, MIT. The vendor's own card states Ornith 1.5 extends Ornith 1.0, which was built on Qwen3.5 and Gemma4 with further continued pretraining, mid-training and post-training - a declared derivative lineage, not a from-scratch model (model card read 2026-08-30). The ornith-1.5:9b tag was verified against the Ollama registry manifest, which serves a 5.63 GB Q4_K_M plus a 0.92 GB vision projector (checked 2026-08-30). Measured on this project's own hardware: 5.5 GB resident at 100% GPU on an RTX 4070 Ti (12 GB), 61.3 tok/s generation (2026-08-21, docs/lab/radar-2026-08-21-model-validation.md).
Ornith AI · Ornith
Model overview
Ornith 1.5 9B is the compact member of the Ornith 1.5 family, a dense 9.65B model on the Qwen3.5 architecture with a 256K context window and image input. Its own model card describes the line as a derivative of Qwen3.5 and Gemma4 carried forward through additional training rather than a model trained from scratch. It is the lightest Ornith release this project measured running entirely inside a 12 GB card at Q4_K_M.
Editorial review
VRAM figures are empirical estimates. Actual usage varies by runtime, context length, and system configuration. Verify on your specific hardware before production use.
Will Ornith 1.5 9B run on your machine?
Ornith 1.5 9B is 9.7B parameters and needs 7.1 GB of VRAM at Q4_K_M — 5.6 GB of weights plus 1.5 GB of runtime overhead for the inference server itself.
VRAM by quantization
| Quantization | Weights | Needs (with overhead) | Quality |
|---|---|---|---|
| Q4_K_M | 5.6 GB | 7.1 GB | good |
Fit on common hardware at Q4_K_M
| Hardware | Memory the model can use | System RAM | Verdict |
|---|---|---|---|
| CPU Only | None (CPU only) | 16 GB | CPU offload |
| RTX 4060 Laptop | 8 GB | 16 GB | Tight |
| RTX 3060 (12GB) | 12 GB | 32 GB | Comfortable |
| RTX 4060 Ti (16GB) | 16 GB | 32 GB | Comfortable |
| RTX 3090 | 24 GB | 64 GB | Comfortable |
Comfortable means VRAM clears the requirement by 2 GB or more. Tight means it covers the requirement with no margin. CPU offload means the model does not fit in VRAM but system RAM is at least 1.6× the weights, so it will run at reduced speed — expect roughly 1–5 tokens per second. Figures are weights plus a fixed runtime overhead and exclude KV-cache growth, which scales with context length.
Need more hardware for Ornith 1.5 9B? Open the PC Builder for the 13B / 14B tier →
Ready to run this model locally?
Find a compatible interface in our Local AI Tools directory →