| Model | Verdict | Decode tok/s | Max ctx | Size | Chat | Coding | |
|---|---|---|---|---|---|---|---|
| Qwen/Qwen3.5-9B unsloth/Qwen3.5-9B-GGUF · Q5_K_M | Fits | 7.302.90–8.00 | 36K | 9.7B | p46 | – | |
| ornith-ai/Ornith-1.5-9B AtomicChat/Ornith-1.5-9B-GGUF · Q5_K_M | Fits | 7.403.00–8.10 | 40K | 9.7B | p46 | p24 | |
| Qwen/Qwen3.5-4B unsloth/Qwen3.5-4B-GGUF · Q8_0 | Fits | 9.003.60–9.90 | 77K | 4.7B | p39 | – | |
| google/gemma-4-12B-it unsloth/gemma-4-12b-it-GGUF · IQ4_NL | Fits | 4.001.60–4.40 | 33K | 12B | p38 | – | |
| Qwen/Qwen3-4B-Instruct-2507 unsloth/Qwen3-4B-Instruct-2507-GGUF · Q8_0 | Fits | 6.202.50–6.90 | 38K | 4B | p38 | – | |
| ornith-ai/Ornith-1.0-9B unsloth/Ornith-1.0-9B-GGUF · Q5_K_M | Fits | 7.302.90–8.10 | 37K | 9.4B | – | p20 | |
| ornith-ai/Ornith-1.5-9B-GGUF Q5_K_M | Fits | 7.302.90–8.00 | 35K | 9.2B | – | – | |
| prism-ml/Bonsai-27B-mlx-1bit MLX-1BIT | Fits | 6.502.60–7.10 | 37K | 26.4B | – | – | |
| poolside/Laguna-XS.2 poolside-laguna-hackathon/Piscina-XS.2-GGUF · IQ1_M | Tight fit | 25.610.2–28.1 | 0 | 33.4B | – | p25 | |
| Qwen/Qwen3-30B-A3B-Instruct-2507 MaziyarPanahi/Qwen3-30B-A3B-Instruct-2507-GGUF · Q6_K | Unknown | – | – | 30.5B | p85 | – |
Quality = benchmark percentile against all indexed models (p50 = median). Speeds are single-user decode ranges.