Ornith 1.5 (9B, Q4)
Ornith AI's dense 9B Ornith 1.5 model — a compact self-improving agentic coder trained for software engineering, tool use, and reasoning. The 4-bit build offers the practical small-tier balance of quality and memory use, with native tool calling and a 256K context window. MIT licensed; best suited to machines with 12GB+ usable memory.
At a glance
- Parameters
- 9B
- Quantization
- Q4_K_M
- Context window
- 262K tokens
- Approx. size
- 5.6 GB
- Engines
- llama.cpp, MLX
- License
- MIT (open)
- Version
- v1.0.1 , released 2026-08-29
- Family
- qwen
Tuning
Defaults a gezel client applies out of the box, from real evaluation runs against this exact quantization.
- sampling.temperature
0.6- sampling.topP
0.95- sampling.topK
20- sampling.minP
0- sampling.maxTokens
8192- sampling.repetitionPenalty
1- reasoning.thinkingBudget
4096- reasoning.enableThinking
true- profiles
- thinking-general, thinking-coding, thinking-precise, instruct, creative
Model behaviors
Client-side behavior modules the catalog enables for this model (reasoning-tag handling, fabrication detection, prompt shaping).
- reasoning.strip-think-tags
- prompt.private-reasoning-guidance
- prompt.tool-cookbook-condensed
- fabrication.detect-past-tense-no-tools
- turn.ollama-num-predict-bumped
- turn.preamble-folding
- turn.ramble-detection
- mcp.compact-tool-schemas
- mcp.relax-required-fields
- mcp.default-missing-fields
- prompt.retrieval-first
- tools.mlx-grammar
Sources
- llama.cpp (GGUF)
- ornith-ai/Ornith-1.5-9B-GGUF ·
Ornith-1.5-9B-Q4_K_M.gguf· Q4_K_M - MLX
- ornith-ai/Ornith-1.5-9B-MLX-4bit
- Upstream
- https://huggingface.co/ornith-ai/Ornith-1.5-9B
- Manifest
- View on GitHub
Tags
- ornith
- coding
- agentic
- tools
- reasoning
- long-context