| Model | Verdict | Decode tok/s | Max ctx | Size | Chat | Coding | |
|---|---|---|---|---|---|---|---|
| Qwen/Qwen3-30B-A3B-Instruct-2507 BF16 | Fits | 77.831.1–85.6 | 256K | 30.5B | p85 | – | |
| google/gemma-4-31B-it unsloth/gemma-4-31B-it-GGUF · Q8_0 | Fits | 20.816.6–26.0 | 256K | 31.3B | p62 | – | |
| Qwen/Qwen3.5-27B unsloth/Qwen3.5-27B-GGUF · Q8_0 | Fits | 26.822.8–30.8 | 256K | 27.8B | p60 | p46 | |
| Qwen/Qwen3.8-27B unsloth/Qwen3.8-27B-GGUF · Q8_0 | Fits | 26.422.5–30.4 | 256K | 27.8B | p58 | p31 | |
| Qwen/Qwen3.6-27B unsloth/Qwen3.6-27B-GGUF · Q8_0 | Fits | 26.822.8–30.8 | 256K | 27.8B | p50 | p54 | |
| Qwen/Qwen3.5-35B-A3B BF16 | Fits | 77.030.8–84.7 | 256K | 36B | p48 | p35 | |
| google/gemma-4-26B-A4B-it BF16 | Fits | 76.830.7–84.5 | 256K | 25.8B | p48 | – | |
| Qwen/Qwen3.5-9B unsloth/Qwen3.5-9B-GGUF · Q8_0 | Fits | 79.167.3–91.0 | 256K | 9.7B | p46 | – | |
| ornith-ai/Ornith-1.5-9B AtomicChat/Ornith-1.5-9B-GGUF · Q8_0 | Fits | 79.167.3–91.0 | 256K | 9.7B | p44 | p23 | |
| Qwen/Qwen3.6-35B-A3B BF16 | Fits | 77.030.8–84.7 | 256K | 36B | p43 | p41 | |
| openbmb/MiniCPM-V-4.7-35B-A3B BF16 | Fits | 88.535.4–97.4 | 256K | 35.2B | p42 | – | |
| openai/gpt-oss-20b MXFP4 | Fits | 13051.9–143 | 128K | 21.1B | p39 | p15 | |
| Qwen/Qwen3.5-4B BF16 | Fits | 74.529.8–81.9 | 256K | 4.7B | p39 | – | |
| ornith-ai/Ornith-1.0-35B BF16 | Fits | 90.536.2–99.5 | 256K | 35.1B | – | p38 | |
| google/gemma-4-12B-it unsloth/gemma-4-12b-it-GGUF · Q8_0 | Fits | 48.341.1–55.6 | 256K | 12B | p38 | – | |
| Qwen/Qwen3-4B-Instruct-2507 BF16 | Fits | 74.329.7–81.7 | 256K | 4B | p38 | – | |
| nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16 BF16 | Fits | 14256.8–156 | 256K | 32.9B | p37 | p8 | |
| nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4 MIXED_PRECISION | Fits | 16766.7–184 | 1M | 47.1B | p36 | p8 | |
| nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16 BF16 | Fits | 14558.0–159 | 256K | 31.6B | p33 | – | |
| poolside/Laguna-XS-2.1 BF16 | Fits | 10140.4–111 | 256K | 33.4B | – | p30 | |
| poolside/Laguna-XS.2 BF16 | Fits | 10140.4–111 | 256K | 33.4B | – | p25 | |
| ornith-ai/Ornith-1.0-9B unsloth/Ornith-1.0-9B-GGUF · Q8_0 | Fits | 79.367.4–91.2 | 256K | 9.4B | – | p19 | |
| ornith-ai/Ornith-1.5-35B-A3B-NVFP4 MIXED_PRECISION | Fits | 19678.4–216 | 256K | 34B | – | – | |
| nvidia/Qwen3.6-35B-A3B-NVFP4 MIXED_PRECISION | Fits | 19678.4–216 | 256K | 34B | – | – | |
| Qwen/Qwen3-Coder-30B-A3B-Instruct-FP8 FP8 | Fits | 10642.4–117 | 256K | 30.5B | – | – | |
| prism-ml/Bonsai-27B-mlx-1bit MLX-1BIT | Fits | 88.935.6–97.8 | 256K | 26.4B | – | – | |
| ornith-ai/Ornith-1.0-35B-GGUF Q8_0 | Fits | 84.667.7–106 | 256K | 34.7B | – | – | |
| ornith-ai/Ornith-1.5-35B-A3B-GGUF Q8_0 | Fits | 83.066.4–104 | 256K | 35.5B | – | – | |
| Qwen/Qwen3-30B-A3B BF16 | Fits | 77.831.1–85.6 | 40K | 30.5B | – | – | |
| Qwen/Qwen3-Coder-30B-A3B-Instruct BF16 | Fits | 77.831.1–85.6 | 256K | 30.5B | – | – | |
| ornith-ai/Ornith-1.5-9B-GGUF Q8_0 | Fits | 77.565.9–89.1 | 256K | 9.2B | – | – | |
| prism-ml/Ternary-Bonsai-27B-mlx-2bit MLX-2BIT | Fits | 67.026.8–73.7 | 256K | 26.4B | – | – | |
| Qwen/Qwen3-14B unsloth/Qwen3-14B-GGUF · Q8_0 | Fits | 40.532.4–50.7 | 40K | 14.8B | – | – | |
| Qwen/Qwen2.5-14B-Instruct Qwen/Qwen2.5-14B-Instruct-GGUF · Q8_0 | Fits | 38.831.1–48.5 | 32K | 14.8B | – | – | |
| Qwen/Qwen2.5-Coder-14B-Instruct Qwen/Qwen2.5-Coder-14B-Instruct-GGUF · Q4_K_M | Fits | 35.028.0–43.8 | 32K | 14.8B | – | – | |
| RadixArk/Qwen3.8-27B-NVFP4 MIXED_PRECISION | Fits | 33.813.5–37.2 | 256K | 26.4B | – | – | |
| farbodtavakkoli/OTel-2.0-LLM-31B-IT wfakhri/OTel-2.0-LLM-31B-IT-GGUF · Q8_0 | Fits | 20.816.6–26.0 | 256K | 31.3B | – | – | |
| Qwen/Qwen2.5-Coder-32B-Instruct Qwen/Qwen2.5-Coder-32B-Instruct-GGUF · Q3_K_M | Fits | 20.416.3–25.5 | 32K | 32.8B | – | – | |
| Qwen/Qwen2.5-32B-Instruct Qwen/Qwen2.5-32B-Instruct-GGUF · Q8_0 | Fits | 19.015.2–23.7 | 32K | 32.8B | – | – | |
| Qwen/Qwen3-32B unsloth/Qwen3-32B-GGUF · Q8_0 | Fits | 18.815.0–23.5 | 40K | 32.8B | – | – |
Quality = benchmark percentile against all indexed models (p50 = median). Speeds are single-user decode ranges.