Ornith 1.0 9B

weights

open(MIT)

base

Qwen3.5

data

closed

recipe

closed

license

MIT(OSI)

MIT-licensed open weights on a Qwen 3.5 base, but no open training data or recipe, so open weights rather than open source.

• https://huggingface.co/api/models/deepreinforce-ai/Ornith-1.0-9B recorded 2026-09-29

  API record for ornith-ai/Ornith-1.0-9B, served for the declared deepreinforce-ai path: 963,330 downloads over the trailing 30 days; tags include license:mit, qwen3_5 and safetensors; the record also carries the sample prompt "Explain quantum computing in simple terms."

• https://huggingface.co/ornith-ai/Ornith-1.0-9B recorded 2026-09-29

  Model page: "MIT licensed, globally accessible, and free from regional limitations." with the Qwen3_5ForConditionalGeneration architecture; searched the body for training data, dataset and recipe links: none.

• https://huggingface.co/ornith-ai/Ornith-1.0-9B/raw/main/README.md recorded 2026-09-29

  Raw model card README: "MIT licensed, globally accessible, and free from regional limitations." and "post-trained on top of Gemma 4 and Qwen 3.5". Searched the body for training data, dataset, recipe, technical report and code repository links: none.

Adoption is measured as Hugging Face downloads summed across Ornith's two declared repositories, the base model and its GGUF build.

• https://huggingface.co/api/models/deepreinforce-ai/Ornith-1.0-9B recorded 2026-09-29

  API record for ornith-ai/Ornith-1.0-9B, served for the declared deepreinforce-ai path: 963,330 downloads over the trailing 30 days; tags include license:mit, qwen3_5 and safetensors; the record also carries the sample prompt "Explain quantum computing in simple terms."

• https://huggingface.co/api/models/deepreinforce-ai/Ornith-1.0-9B-GGUF recorded 2026-09-29

  API record for ornith-ai/Ornith-1.0-9B-GGUF, served for the declared deepreinforce-ai path: 2,666,462 downloads over the trailing 30 days; the record also carries the sample prompt "Explain quantum computing in simple terms."

Ornith 1.0 9B is vendor-reported at 69.4 on SWE-bench Verified, 42.9 on SWE-bench Pro and 43.1 on Terminal-Bench 2.1 with Terminus-2, all from its own model card. Those results are strong for a 9B model but sit below the frontier coding models.

• https://huggingface.co/ornith-ai/Ornith-1.0-9B/raw/main/README.md recorded 2026-09-29

  Model card results table: Terminal-Bench 2.1 with Terminus-2 at 43.1 and Claude Code at 40.6, SWE-bench Verified 69.4, SWE-bench Pro 42.9, SWE-bench Multilingual 52, Claw-eval Avg 63.1; the card claims "achieving state-of-the-art performance among open-source models of comparable size" and describes a reasoning model with tool-call parsing.