0
Quantized builds
98
Benchmark records
1
Discussions
Quantized builds
Community and official quantized releases; "linked records" counts only benchmarks tied to that exact release.
No quantized builds cataloged yet.
Benchmark data
50 configurations
| Hardware | Framework | Quant | Decode | VRAM used | Samples | Evidence |
|---|---|---|---|---|---|---|
| NVIDIA RTX Spark | llama.cpp | Q4_K_M | 53.5 tok/s | 15.88 GB | 2 | L2 Cross-framework verified |
| NVIDIA RTX Spark | llama.cpp | Q5_K_M | 46.7 tok/s | 17.72 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX Spark | llama.cpp | Q8_0 | 35.4 tok/s | 21.42 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX Spark | Ollama | FP16 | 27.9 tok/s | 33.18 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX Spark | Ollama | Q6_K | 40.6 tok/s | 18.9 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX Spark | Ollama | Q8_0 | 37 tok/s | 21.42 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX 4090 | vLLM | IQ2_XS | 205.1 tok/s | 10.16 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX 4090 | Strata | Q4_K_M | 191.7 tok/s | 15.88 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX 4090 | Strata | Q6_K | 151.9 tok/s | 18.9 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX 3090 | vLLM | Q8_0 | 64.4 tok/s | 21.42 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX 3090 | Ollama | IQ3_XS | 101 tok/s | 12.52 GB | 1 | L2 Cross-framework verified |
| AMD Radeon RX 7900 XTX | llama.cpp | Q3_K_S | 102.2 tok/s | 13.02 GB | 1 | L2 Cross-framework verified |
| AMD Radeon RX 7900 XTX | llama.cpp | Q4_K_M | 83.5 tok/s | 15.88 GB | 1 | L2 Cross-framework verified |
| AMD Radeon RX 7900 XTX | vLLM | Q3_K_M | 114.5 tok/s | 13.52 GB | 1 | L2 Cross-framework verified |
| AMD Radeon RX 7900 XTX | vLLM | Q4_K_M | 87.1 tok/s | 15.88 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX 5090 | Ollama | Q5_K_S | 220.5 tok/s | 17.22 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX 5090 | Strata | IQ3_S | 280 tok/s | 13.19 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX 5090 | Strata | Q5_K_M | 298 tok/s | 17.72 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX 5080 | Ollama | Q5_K_S | 149.5 tok/s | 15.2 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX 5080 | Strata | IQ3_S | 303 tok/s | 13.19 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX 3060 12GB | llama.cpp | IQ3_S | 72.5 tok/s | 11.4 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX 3060 12GB | vLLM | IQ2_XS | 94.6 tok/s | 10.16 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX 4080 Super | llama.cpp | Q3_K_M | 201.4 tok/s | 13.52 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX 4080 Super | Strata | IQ2_XS | 307.8 tok/s | 10.16 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX 4080 Super | Strata | Q5_K_M | 179.7 tok/s | 15.2 GB | 1 | L2 Cross-framework verified |
| AMD Ryzen AI Max+ 395 | llama.cpp | IQ3_XS | 45.9 tok/s | 12.52 GB | 1 | L2 Cross-framework verified |
| AMD Ryzen AI Max+ 395 | vLLM | Q2_K | 51.2 tok/s | 11.34 GB | 1 | L2 Cross-framework verified |
| Apple M5 Max | llama.cpp | IQ2_M | 101.5 tok/s | 10.84 GB | 1 | L2 Cross-framework verified |
| Apple M5 Max | llama.cpp | Q3_K_L | 81.2 tok/s | 14.2 GB | 1 | L2 Cross-framework verified |
| Apple M5 Max | Ollama | IQ3_XS | 82.8 tok/s | 12.52 GB | 1 | L2 Cross-framework verified |
| Apple M5 Max | Ollama | Q3_K_S | 79.8 tok/s | 13.02 GB | 1 | L2 Cross-framework verified |
| Apple M5 Ultra | llama.cpp | Q3_K_S | 134.8 tok/s | 13.02 GB | 1 | L2 Cross-framework verified |
| Apple M5 Ultra | llama.cpp | Q5_K_S | 104.8 tok/s | 17.22 GB | 1 | L2 Cross-framework verified |
| Apple M5 Ultra | Ollama | Q8_0 | 65.7 tok/s | 21.42 GB | 1 | L2 Cross-framework verified |
| NVIDIA H100 80GB HBM3 | llama.cpp | IQ1_S | 337.5 tok/s | 9.16 GB | 1 | L2 Cross-framework verified |
| NVIDIA H100 80GB HBM3 | llama.cpp | Q4_K_S | 314.6 tok/s | 15.2 GB | 1 | L2 Cross-framework verified |
| NVIDIA H100 80GB HBM3 | Strata | Q5_K_M | 362.6 tok/s | 17.72 GB | 1 | L2 Cross-framework verified |
| NVIDIA H20 | llama.cpp | Q3_K_M | 65.3 tok/s | 13.52 GB | 1 | L2 Cross-framework verified |
| NVIDIA H20 | Ollama | IQ3_S | 56.6 tok/s | 13.19 GB | 1 | L2 Cross-framework verified |
| NVIDIA H20 | Ollama | Q5_K_M | 43.7 tok/s | 17.72 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX PRO 6000 Blackwell 96GB | Ollama | Q3_K_L | 262.1 tok/s | 14.2 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX PRO 6000 Blackwell 96GB | Ollama | Q5_K_M | 184.1 tok/s | 17.72 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX PRO 6000 Blackwell 96GB | Strata | IQ4_XS | 387.3 tok/s | 14.53 GB | 1 | L2 Cross-framework verified |
| Intel Arc B580 | llama.cpp | IQ3_S | 58.5 tok/s | 11.4 GB | 1 | L2 Cross-framework verified |
| Intel Arc B580 | llama.cpp | Q3_K_S | 65.5 tok/s | 11.4 GB | 1 | L2 Cross-framework verified |
| Intel Arc B580 | Ollama | IQ2_XS | 71 tok/s | 10.16 GB | 1 | L2 Cross-framework verified |
| Apple M2 Ultra | llama.cpp | Q3_K_L | 129.5 tok/s | 14.2 GB | 1 | L2 Cross-framework verified |
| Apple M2 Ultra | llama.cpp | Q4_K_S | 123.2 tok/s | 15.2 GB | 1 | L2 Cross-framework verified |
| Apple M2 Ultra | Ollama | Q4_K_S | 103.2 tok/s | 15.2 GB | 1 | L2 Cross-framework verified |
| NVIDIA RTX Spark | llama.cpp | IQ1_S | 80.8 tok/s | 9.16 GB | 1 | L2 Cross-framework verified |
98 of these records are not linked to a specific quantized build; the full set is on the search page.