{"data":{"accepted_at":null,"id":"qwen3-8-27b-gptq-int4-intel-arc-pro-b70-32gb-vllm-tp2-sweep","measured_at":"2026-08-21T10:14:21.225Z","metrics":{"concurrency":null,"inference_engine_version":"vLLM 0.26.1rc1.dev771+g8e6d8e4f6 XPU (editable tree, custom-patched: host-staged TP2 collectives + load-time INT8 W8A8 lm_head via oneDNN kernels)","latest_point_at":"2026-08-21T10:14:21.225Z","max_context_tokens":2048,"peak_generation_tps":105.6,"peak_prompt_tps":710.3,"point_count":1},"recipe_id":"qwen3-8-27b-gptq-int4-intel-arc-pro-b70-32gb-vllm-tp2","rows":[{"concurrency":null,"context_tokens":2048,"decode_tok_s":105.6,"decode_tok_s_per_stream":null,"output_tokens":102,"peak_vram_gb":null,"prefill_tok_s":710.3,"samples":1,"status":"observed","ttft_ms_p50":891.22}],"schema_version":"local-ai-registry/v1","source":{"commit":null,"kind":"leaderboard","paths":["/en/runs/cmt2slgyx0hdgmv01d8vf53cr"],"repository":"https://www.localmaxxing.com","url":"https://www.localmaxxing.com/en/runs/cmt2slgyx0hdgmv01d8vf53cr"},"relationships":{"recipe":{"api":"/api/v1/recipes/qwen3-8-27b-gptq-int4-intel-arc-pro-b70-32gb-vllm-tp2","href":"/recipes/qwen3-8-27b-gptq-int4-intel-arc-pro-b70-32gb-vllm-tp2","id":"qwen3-8-27b-gptq-int4-intel-arc-pro-b70-32gb-vllm-tp2"}}},"meta":{"source":"registry"}}