{"schema_version":2,"id":"qwen27-autoround-115k-c1","status":"pass","mode":"vllm","model":{"name":"Qwen3.6-27B-int4-AutoRound","bytes":18996705768,"parameters_b":27.0,"active_parameters_b":null,"quant":"safetensors","family":"Qwen","moe":false},"context":131072,"workload":{"pp":2048,"tg":512,"depth":115404,"target_tokens":117964,"target_fill_pct":90.0,"actual_fill_pct":89.9994,"runs":1,"warmup_runs":1,"concurrency":1},"settings":{"kv":"auto","k_cache_dtype":"auto","v_cache_dtype":"auto","kv_cache_label":"Auto (model dtype)","attention_backend":"auto","gpu_memory":0.9,"max_num_seqs":null,"max_num_batched_tokens":8192,"logical_batch_tokens":null,"physical_batch_tokens":null,"execution_mode":"compiled_xpu_graph_full_decode_only","engine":"vLLM XPU","llama_backend":null,"full_context_kernel_warmup":false,"jit_warmup_retried":true,"speculative_method":null,"mtp_tokens":null},"measurement":{"load_seconds":108.051,"pp":1024.5439992345373,"tg":21.727573166473405,"first_response_ms":114707.91524299784,"benchmark_seconds":142.655,"seconds":404.7827434539795}}
{"schema_version":2,"id":"qwen27-q4ks-115k-c1","status":"pass","mode":"llama","model":{"name":"Qwen3.6-27B-Q4_K_S.gguf","bytes":16121357440,"parameters_b":27.0,"active_parameters_b":null,"quant":"Q4_K_S","family":"Qwen","moe":false},"context":131072,"workload":{"pp":2048,"tg":512,"depth":115404,"target_tokens":117964,"target_fill_pct":90.0,"actual_fill_pct":89.9994,"runs":1,"warmup_runs":0,"concurrency":1},"settings":{"kv":"f16","k_cache_dtype":null,"v_cache_dtype":null,"kv_cache_label":null,"attention_backend":"auto","gpu_memory":null,"max_num_seqs":null,"max_num_batched_tokens":null,"logical_batch_tokens":4096,"physical_batch_tokens":4096,"execution_mode":null,"engine":"llama.cpp SYCL","llama_backend":"sycl","full_context_kernel_warmup":false,"jit_warmup_retried":false,"speculative_method":null,"mtp_tokens":null},"measurement":{"load_seconds":52.303,"pp":130.36187498075375,"tg":10.6861226909714,"first_response_ms":901216.7649900003,"benchmark_seconds":953.009,"seconds":1005.3411302566528}}
{"schema_version":2,"id":"qwen35a3b-q4ks-115k-c1","status":"pass","mode":"llama","model":{"name":"Qwen3.6-35B-A3B-UD-Q4_K_S.gguf","bytes":21388319008,"parameters_b":35.0,"active_parameters_b":3.0,"quant":"UD-Q4_K_S","family":"Qwen","moe":true},"context":131072,"workload":{"pp":2048,"tg":512,"depth":115404,"target_tokens":117964,"target_fill_pct":90.0,"actual_fill_pct":89.9994,"runs":1,"warmup_runs":0,"concurrency":1},"settings":{"kv":"f16","k_cache_dtype":null,"v_cache_dtype":null,"kv_cache_label":null,"attention_backend":"auto","gpu_memory":null,"max_num_seqs":null,"max_num_batched_tokens":null,"logical_batch_tokens":4096,"physical_batch_tokens":4096,"execution_mode":null,"engine":"llama.cpp SYCL","llama_backend":"sycl","full_context_kernel_warmup":false,"jit_warmup_retried":false,"speculative_method":null,"mtp_tokens":null},"measurement":{"load_seconds":59.299,"pp":528.0353750975698,"tg":42.634127448022554,"first_response_ms":222529.55932799887,"benchmark_seconds":237.328,"seconds":296.63581013679504}}
{"schema_version":2,"id":"qwen35a3b-autoround-115k-c1","status":"pass","mode":"vllm","model":{"name":"Qwen3.6-35B-A3B-int4-mixed-AutoRound","bytes":21494033400,"parameters_b":35.0,"active_parameters_b":3.0,"quant":"safetensors","family":"Qwen","moe":true},"context":131072,"workload":{"pp":2048,"tg":512,"depth":115404,"target_tokens":117964,"target_fill_pct":90.0,"actual_fill_pct":89.9994,"runs":1,"warmup_runs":0,"concurrency":1},"settings":{"kv":"auto","k_cache_dtype":"auto","v_cache_dtype":"auto","kv_cache_label":"Auto (model dtype)","attention_backend":"auto","gpu_memory":0.9,"max_num_seqs":null,"max_num_batched_tokens":8192,"logical_batch_tokens":null,"physical_batch_tokens":null,"execution_mode":"compiled_xpu_graph_full_decode_only","engine":"vLLM XPU","llama_backend":null,"full_context_kernel_warmup":false,"jit_warmup_retried":false,"speculative_method":null,"mtp_tokens":null},"measurement":{"load_seconds":185.068,"pp":3377.1338682596543,"tg":47.030922104512875,"first_response_ms":34904.502377001336,"benchmark_seconds":50.784,"seconds":239.31531691551208}}
{"schema_version":2,"id":"ornith35-autoround-115k-c1","status":"pass","mode":"vllm","model":{"name":"Ornith-1.0-35B-int4-AutoRound","bytes":20451906392,"parameters_b":35.0,"active_parameters_b":null,"quant":"safetensors","family":"Qwen","moe":true},"context":131072,"workload":{"pp":2048,"tg":512,"depth":115404,"target_tokens":117964,"target_fill_pct":90.0,"actual_fill_pct":89.9994,"runs":1,"warmup_runs":0,"concurrency":1},"settings":{"kv":"auto","k_cache_dtype":"auto","v_cache_dtype":"auto","kv_cache_label":"Auto (model dtype)","attention_backend":"auto","gpu_memory":0.9,"max_num_seqs":null,"max_num_batched_tokens":8192,"logical_batch_tokens":null,"physical_batch_tokens":null,"execution_mode":"compiled_xpu_graph_full_decode_only","engine":"vLLM XPU","llama_backend":null,"full_context_kernel_warmup":false,"jit_warmup_retried":false,"speculative_method":null,"mtp_tokens":null},"measurement":{"load_seconds":182.065,"pp":3312.7590610340562,"tg":47.8268101133891,"first_response_ms":35585.45378399867,"benchmark_seconds":51.347,"seconds":238.7174894809723}}
{"schema_version":2,"id":"ornith35-q4ks-115k-c1","status":"pass","mode":"llama","model":{"name":"Ornith-1.0-35B-UD-Q4_K_S.gguf","bytes":20893014432,"parameters_b":35.0,"active_parameters_b":null,"quant":"UD-Q4_K_S","family":"Qwen","moe":true},"context":131072,"workload":{"pp":2048,"tg":512,"depth":115404,"target_tokens":117964,"target_fill_pct":90.0,"actual_fill_pct":89.9994,"runs":1,"warmup_runs":0,"concurrency":1},"settings":{"kv":"f16","k_cache_dtype":null,"v_cache_dtype":null,"kv_cache_label":null,"attention_backend":"auto","gpu_memory":null,"max_num_seqs":null,"max_num_batched_tokens":null,"logical_batch_tokens":4096,"physical_batch_tokens":4096,"execution_mode":null,"engine":"llama.cpp SYCL","llama_backend":"sycl","full_context_kernel_warmup":false,"jit_warmup_retried":false,"speculative_method":null,"mtp_tokens":null},"measurement":{"load_seconds":58.3,"pp":526.0913387647867,"tg":42.690472377392005,"first_response_ms":223349.484579001,"benchmark_seconds":238.237,"seconds":296.5510776042938}}
{"schema_version":2,"id":"gemma12-autoround-4k-c1","status":"pass","mode":"vllm","model":{"name":"gemma-4-12B-it-int4-AutoRound","bytes":7782599832,"parameters_b":12.0,"active_parameters_b":null,"quant":"safetensors","family":"Gemma","moe":false},"context":4096,"workload":{"pp":2048,"tg":512,"depth":1126,"target_tokens":3686,"target_fill_pct":90.0,"actual_fill_pct":89.9902,"runs":1,"warmup_runs":0,"concurrency":1},"settings":{"kv":"auto","k_cache_dtype":"auto","v_cache_dtype":"auto","kv_cache_label":"Auto (model dtype)","attention_backend":"auto","gpu_memory":0.9,"max_num_seqs":1,"max_num_batched_tokens":4096,"logical_batch_tokens":null,"physical_batch_tokens":null,"execution_mode":"compiled_xpu_graph_full_decode_only","engine":"vLLM XPU","llama_backend":null,"full_context_kernel_warmup":false,"jit_warmup_retried":false,"speculative_method":"none","mtp_tokens":0},"measurement":{"load_seconds":73.043,"pp":3363.8451386985307,"tg":51.855013329132866,"first_response_ms":981.4488319971133,"benchmark_seconds":15.793,"seconds":90.67726445198059}}
{"schema_version":2,"id":"gemma12-bf16-4k-c1","status":"pass","mode":"vllm","model":{"name":"gemma-4-12B-it","bytes":23919549408,"parameters_b":12.0,"active_parameters_b":null,"quant":"safetensors","family":"Gemma","moe":false},"context":4096,"workload":{"pp":2048,"tg":512,"depth":1126,"target_tokens":3686,"target_fill_pct":90.0,"actual_fill_pct":89.9902,"runs":1,"warmup_runs":0,"concurrency":1},"settings":{"kv":"auto","k_cache_dtype":"auto","v_cache_dtype":"auto","kv_cache_label":"Auto (model dtype)","attention_backend":"auto","gpu_memory":0.9,"max_num_seqs":1,"max_num_batched_tokens":4096,"logical_batch_tokens":null,"physical_batch_tokens":null,"execution_mode":"compiled_xpu_graph_full_decode_only","engine":"vLLM XPU","llama_backend":null,"full_context_kernel_warmup":false,"jit_warmup_retried":false,"speculative_method":"none","mtp_tokens":0},"measurement":{"load_seconds":84.065,"pp":4420.181435616955,"tg":21.261593598940227,"first_response_ms":783.0495739999606,"benchmark_seconds":30.267,"seconds":116.49312734603882}}
{"schema_version":2,"id":"gemma12-w4a16-4k-c1","status":"pass","mode":"vllm","model":{"name":"gemma-4-12B-it-qat-w4a16","bytes":10264229896,"parameters_b":12.0,"active_parameters_b":null,"quant":"safetensors","family":"Gemma","moe":false},"context":4096,"workload":{"pp":2048,"tg":512,"depth":1126,"target_tokens":3686,"target_fill_pct":90.0,"actual_fill_pct":89.9902,"runs":1,"warmup_runs":0,"concurrency":1},"settings":{"kv":"auto","k_cache_dtype":"auto","v_cache_dtype":"auto","kv_cache_label":"Auto (model dtype)","attention_backend":"auto","gpu_memory":0.9,"max_num_seqs":1,"max_num_batched_tokens":4096,"logical_batch_tokens":null,"physical_batch_tokens":null,"execution_mode":"compiled_xpu_graph_full_decode_only","engine":"vLLM XPU","llama_backend":null,"full_context_kernel_warmup":false,"jit_warmup_retried":false,"speculative_method":"none","mtp_tokens":0},"measurement":{"load_seconds":77.045,"pp":628.0142621283293,"tg":19.267327567259844,"first_response_ms":5103.021460000036,"benchmark_seconds":51.11,"seconds":130.01033973693848}}
{"schema_version":2,"id":"gemma12-autoround-8k-b8192-c1","status":"pass","mode":"vllm","model":{"name":"gemma-4-12B-it-int4-AutoRound","bytes":7782599832,"parameters_b":12.0,"active_parameters_b":null,"quant":"safetensors","family":"Gemma","moe":false},"context":8192,"workload":{"pp":2048,"tg":512,"depth":4812,"target_tokens":7372,"target_fill_pct":90.0,"actual_fill_pct":89.9902,"runs":1,"warmup_runs":0,"concurrency":1},"settings":{"kv":"auto","k_cache_dtype":"auto","v_cache_dtype":"auto","kv_cache_label":"Auto (model dtype)","attention_backend":"auto","gpu_memory":0.9,"max_num_seqs":1,"max_num_batched_tokens":8192,"logical_batch_tokens":null,"physical_batch_tokens":null,"execution_mode":"compiled_xpu_graph_full_decode_only","engine":"vLLM XPU","llama_backend":null,"full_context_kernel_warmup":false,"jit_warmup_retried":false,"speculative_method":"none","mtp_tokens":0},"measurement":{"load_seconds":65.039,"pp":2862.459266570642,"tg":45.43162776450175,"first_response_ms":2433.8250439905096,"benchmark_seconds":18.655,"seconds":85.6047830581665}}
{"schema_version":2,"id":"gemma12-autoround-8k-b8192-c2","status":"pass","mode":"vllm","model":{"name":"gemma-4-12B-it-int4-AutoRound","bytes":7782599832,"parameters_b":12.0,"active_parameters_b":null,"quant":"safetensors","family":"Gemma","moe":false},"context":8192,"workload":{"pp":2048,"tg":512,"depth":4812,"target_tokens":7372,"target_fill_pct":90.0,"actual_fill_pct":89.9902,"runs":1,"warmup_runs":0,"concurrency":2},"settings":{"kv":"auto","k_cache_dtype":"auto","v_cache_dtype":"auto","kv_cache_label":"Auto (model dtype)","attention_backend":"auto","gpu_memory":0.9,"max_num_seqs":2,"max_num_batched_tokens":8192,"logical_batch_tokens":null,"physical_batch_tokens":null,"execution_mode":"compiled_xpu_graph_full_decode_only","engine":"vLLM XPU","llama_backend":null,"full_context_kernel_warmup":false,"jit_warmup_retried":false,"speculative_method":"none","mtp_tokens":0},"measurement":{"load_seconds":75.047,"pp":2812.3793808453615,"tg":73.46109361887895,"first_response_ms":3907.2711400003755,"benchmark_seconds":21.76,"seconds":98.79706144332886}}
{"schema_version":2,"id":"gemma12-autoround-8k-b8192-c4","status":"pass","mode":"vllm","model":{"name":"gemma-4-12B-it-int4-AutoRound","bytes":7782599832,"parameters_b":12.0,"active_parameters_b":null,"quant":"safetensors","family":"Gemma","moe":false},"context":8192,"workload":{"pp":2048,"tg":512,"depth":4812,"target_tokens":7372,"target_fill_pct":90.0,"actual_fill_pct":89.9902,"runs":1,"warmup_runs":0,"concurrency":4},"settings":{"kv":"auto","k_cache_dtype":"auto","v_cache_dtype":"auto","kv_cache_label":"Auto (model dtype)","attention_backend":"auto","gpu_memory":0.9,"max_num_seqs":4,"max_num_batched_tokens":8192,"logical_batch_tokens":null,"physical_batch_tokens":null,"execution_mode":"compiled_xpu_graph_full_decode_only","engine":"vLLM XPU","llama_backend":null,"full_context_kernel_warmup":false,"jit_warmup_retried":false,"speculative_method":"none","mtp_tokens":0},"measurement":{"load_seconds":67.041,"pp":2928.9295525855664,"tg":109.5624341981165,"first_response_ms":6576.761733747844,"benchmark_seconds":26.551,"seconds":95.5521559715271}}
{"schema_version":2,"id":"gemma12-bf16-8k-c1","status":"pass","mode":"vllm","model":{"name":"gemma-4-12B-it","bytes":23919549408,"parameters_b":12.0,"active_parameters_b":null,"quant":"safetensors","family":"Gemma","moe":false},"context":8192,"workload":{"pp":2048,"tg":512,"depth":4812,"target_tokens":7372,"target_fill_pct":90.0,"actual_fill_pct":89.9902,"runs":1,"warmup_runs":0,"concurrency":1},"settings":{"kv":"auto","k_cache_dtype":"auto","v_cache_dtype":"auto","kv_cache_label":"Auto (model dtype)","attention_backend":"auto","gpu_memory":0.98,"max_num_seqs":1,"max_num_batched_tokens":8192,"logical_batch_tokens":null,"physical_batch_tokens":null,"execution_mode":"compiled_xpu_graph_full_decode_only","engine":"vLLM XPU","llama_backend":null,"full_context_kernel_warmup":false,"jit_warmup_retried":false,"speculative_method":"none","mtp_tokens":0},"measurement":{"load_seconds":88.054,"pp":3656.7882977357094,"tg":20.079928575368427,"first_response_ms":1940.322279999691,"benchmark_seconds":32.763,"seconds":123.15927195549011}}
{"schema_version":2,"id":"gemma12-bf16-8k-c4","status":"pass","mode":"vllm","model":{"name":"gemma-4-12B-it","bytes":23919549408,"parameters_b":12.0,"active_parameters_b":null,"quant":"safetensors","family":"Gemma","moe":false},"context":8192,"workload":{"pp":2048,"tg":512,"depth":4812,"target_tokens":7372,"target_fill_pct":90.0,"actual_fill_pct":89.9902,"runs":1,"warmup_runs":0,"concurrency":4},"settings":{"kv":"auto","k_cache_dtype":"auto","v_cache_dtype":"auto","kv_cache_label":"Auto (model dtype)","attention_backend":"auto","gpu_memory":0.98,"max_num_seqs":4,"max_num_batched_tokens":8192,"logical_batch_tokens":null,"physical_batch_tokens":null,"execution_mode":"compiled_xpu_graph_full_decode_only","engine":"vLLM XPU","llama_backend":null,"full_context_kernel_warmup":false,"jit_warmup_retried":false,"speculative_method":"none","mtp_tokens":0},"measurement":{"load_seconds":67.038,"pp":3515.3482144824347,"tg":64.85150773875704,"first_response_ms":5497.685356000147,"benchmark_seconds":39.236,"seconds":108.13260698318481}}
{"schema_version":2,"id":"gemma12-autoround-115k-c1","status":"pass","mode":"vllm","model":{"name":"gemma-4-12B-it-int4-AutoRound","bytes":7782599832,"parameters_b":12.0,"active_parameters_b":null,"quant":"safetensors","family":"Gemma","moe":false},"context":131072,"workload":{"pp":2048,"tg":512,"depth":115404,"target_tokens":117964,"target_fill_pct":90.0,"actual_fill_pct":89.9994,"runs":1,"warmup_runs":0,"concurrency":1},"settings":{"kv":"auto","k_cache_dtype":"auto","v_cache_dtype":"auto","kv_cache_label":"Auto (model dtype)","attention_backend":"auto","gpu_memory":0.9,"max_num_seqs":1,"max_num_batched_tokens":4096,"logical_batch_tokens":null,"physical_batch_tokens":null,"execution_mode":"compiled_xpu_graph_full_decode_only","engine":"vLLM XPU","llama_backend":null,"full_context_kernel_warmup":false,"jit_warmup_retried":false,"speculative_method":"none","mtp_tokens":0},"measurement":{"load_seconds":70.049,"pp":1335.8555664444582,"tg":16.750590908844526,"first_response_ms":87960.25588200428,"benchmark_seconds":123.688,"seconds":195.63130450248718}}
{"schema_version":2,"id":"gemma12-bf16-115k-c1","status":"pass","mode":"vllm","model":{"name":"gemma-4-12B-it","bytes":23919549408,"parameters_b":12.0,"active_parameters_b":null,"quant":"safetensors","family":"Gemma","moe":false},"context":131072,"workload":{"pp":2048,"tg":512,"depth":115404,"target_tokens":117964,"target_fill_pct":90.0,"actual_fill_pct":89.9994,"runs":1,"warmup_runs":0,"concurrency":1},"settings":{"kv":"auto","k_cache_dtype":"auto","v_cache_dtype":"auto","kv_cache_label":"Auto (model dtype)","attention_backend":"auto","gpu_memory":0.98,"max_num_seqs":1,"max_num_batched_tokens":8192,"logical_batch_tokens":null,"physical_batch_tokens":null,"execution_mode":"compiled_xpu_graph_full_decode_only","engine":"vLLM XPU","llama_backend":null,"full_context_kernel_warmup":false,"jit_warmup_retried":false,"speculative_method":"none","mtp_tokens":0},"measurement":{"load_seconds":27.027,"pp":853.2864132980264,"tg":11.38340546381174,"first_response_ms":137710.7383810003,"benchmark_seconds":184.341,"seconds":213.17806029319763}}
{"schema_version":2,"id":"qwen27-autoround-k4v4-115k-c1","status":"pass","mode":"vllm","model":{"name":"Qwen3.6-27B-int4-AutoRound","bytes":18996705768,"parameters_b":27.0,"active_parameters_b":null,"quant":"safetensors","family":"Qwen","moe":false},"context":131072,"workload":{"pp":2048,"tg":512,"depth":115404,"target_tokens":117964,"target_fill_pct":90.0,"actual_fill_pct":89.9994,"runs":1,"warmup_runs":1,"concurrency":1},"settings":{"kv":"turboquant_4bit_nc","k_cache_dtype":"turboquant_k4_nc","v_cache_dtype":"turboquant_v4","kv_cache_label":"TurboQuant K4 / V4 + norm correction","attention_backend":"TURBOQUANT","gpu_memory":0.9,"max_num_seqs":null,"max_num_batched_tokens":8192,"logical_batch_tokens":null,"physical_batch_tokens":null,"execution_mode":"compiled_xpu_graph_full_decode_only","engine":"vLLM XPU","llama_backend":null,"full_context_kernel_warmup":true,"jit_warmup_retried":false,"speculative_method":null,"mtp_tokens":null},"measurement":{"load_seconds":113.052,"pp":1026.5170997261484,"tg":11.070981723571276,"first_response_ms":114488.57253800088,"benchmark_seconds":165.11,"seconds":409.0614483356476}}
{"schema_version":2,"id":"qwen27-autoround-k8v4-115k-c1","status":"pass","mode":"vllm","model":{"name":"Qwen3.6-27B-int4-AutoRound","bytes":18996705768,"parameters_b":27.0,"active_parameters_b":null,"quant":"safetensors","family":"Qwen","moe":false},"context":131072,"workload":{"pp":2048,"tg":512,"depth":115404,"target_tokens":117964,"target_fill_pct":90.0,"actual_fill_pct":89.9994,"runs":1,"warmup_runs":1,"concurrency":1},"settings":{"kv":"turboquant_k8v4","k_cache_dtype":"turboquant_k8","v_cache_dtype":"turboquant_v4","kv_cache_label":"TurboQuant K8 (FP8) / V4","attention_backend":"TURBOQUANT","gpu_memory":0.9,"max_num_seqs":null,"max_num_batched_tokens":8192,"logical_batch_tokens":null,"physical_batch_tokens":null,"execution_mode":"compiled_xpu_graph_full_decode_only","engine":"vLLM XPU","llama_backend":null,"full_context_kernel_warmup":true,"jit_warmup_retried":false,"speculative_method":null,"mtp_tokens":null},"measurement":{"load_seconds":112.05,"pp":1034.02898860313,"tg":7.500777649177771,"first_response_ms":113658.18135500012,"benchmark_seconds":186.207,"seconds":428.0843107700348}}
{"schema_version":2,"id":"gemma12-autoround-8k-b4096-c1","status":"pass","mode":"vllm","model":{"name":"gemma-4-12B-it-int4-AutoRound","bytes":7782599832,"parameters_b":12.0,"active_parameters_b":null,"quant":"safetensors","family":"Gemma","moe":false},"context":8192,"workload":{"pp":2048,"tg":512,"depth":4812,"target_tokens":7372,"target_fill_pct":90.0,"actual_fill_pct":89.9902,"runs":1,"warmup_runs":0,"concurrency":1},"settings":{"kv":"auto","k_cache_dtype":"auto","v_cache_dtype":"auto","kv_cache_label":"Auto (model dtype)","attention_backend":"auto","gpu_memory":0.9,"max_num_seqs":1,"max_num_batched_tokens":4096,"logical_batch_tokens":null,"physical_batch_tokens":null,"execution_mode":"compiled_xpu_graph_full_decode_only","engine":"vLLM XPU","llama_backend":null,"full_context_kernel_warmup":false,"jit_warmup_retried":false,"speculative_method":"none","mtp_tokens":0},"measurement":{"load_seconds":64.04,"pp":2802.9032498703004,"tg":45.64235448237059,"first_response_ms":2484.6873919887003,"benchmark_seconds":18.653,"seconds":84.65391397476196}}