---

license: mit

base_model:

• ornith-ai/Ornith-1.5-9B

language:

• en

pipeline_tag: image-text-to-text

library_name: transformers

tags:

• text-generation-inference
• llama-cpp
• qwen3_5

---

Ornith-1.5-9B-GGUF

> Ornith-1.5-9B is an open-weight, 9-billion parameter dense reasoning model released under the MIT license by Ornith AI, engineered for single-GPU inference and mobile edge deployment while advancing end-to-end autonomous self-improvement. Built on architectural lineages stemming from Qwen3.5 and Gemma-4 with extensive continued pre-training, mid-training, and reinforcement learning, the model transitions away from static human-curated datasets by operating an internal self-improvement loop that dynamically synthesizes new training tasks, discovers effective execution harnesses, and optimizes solution rollouts. It natively functions as a chain-of-thought reasoner—separating intermediate <think> deliberations from final outputs—while providing structured, XML-parsed tool calling compatible with OpenAI APIs and major agent runtimes like vLLM, SGLang, Ollama, and llama.cpp. With a native context window of 262,144 tokens expandable up to 1M tokens via YaRN RoPE scaling, Ornith-1.5-9B demonstrates remarkable coding, reasoning, and tool-use performance that rivals substantially larger models, achieving high marks across demanding agentic benchmarks including SWE-bench Verified (70.6%), GPQA Diamond (86.4%), Terminal-Bench 2.1, and MCP-Atlas.

Model Files

File Name | Quant Type | File Size | File Link |

|-----------|------------|-----------|-----------|

| Ornith-1.5-9B.BF16.gguf | BF16 | 17.9 GB | Download |

| Ornith-1.5-9B.F16.gguf | F16 | 17.9 GB | Download |

| Ornith-1.5-9B.Q3_K_L.gguf | Q3_K_L | 4.93 GB | Download |

| Ornith-1.5-9B.Q3_K_M.gguf | Q3_K_M | 4.62 GB | Download |

| Ornith-1.5-9B.Q3_K_S.gguf | Q3_K_S | 4.26 GB | Download |

| Ornith-1.5-9B.Q4_0.gguf | Q4_0 | 5.31 GB | Download |

| Ornith-1.5-9B.Q4_K_M.gguf | Q4_K_M | 5.63 GB | Download |

| Ornith-1.5-9B.Q4_K_S.gguf | Q4_K_S | 5.35 GB | Download |

| Ornith-1.5-9B.Q5_0.gguf | Q5_0 | 6.31 GB | Download |

| Ornith-1.5-9B.Q5_K_M.gguf | Q5_K_M | 6.47 GB | Download |

| Ornith-1.5-9B.Q5_K_S.gguf | Q5_K_S | 6.31 GB | Download |

| Ornith-1.5-9B.Q6_K.gguf | Q6_K | 7.36 GB | Download |

| Ornith-1.5-9B.Q8_0.gguf | Q8_0 | 9.53 GB | Download |

| Ornith-1.5-9B.mmproj-bf16.gguf | mmproj-bf16 | 922 MB | Download |

| Ornith-1.5-9B.mmproj-f16.gguf | mmproj-f16 | 922 MB | Download |

| Ornith-1.5-9B.mmproj-q8_0.gguf | mmproj-q8_0 | 624 MB | Download |

llama.cpp

LLM inference in C/C++ — https://github.com/ggml-org/llama.cpp