Speed sweep

qwen3-6-35b-a3b-gptq-4bit-intel-arc-pro-b70-32gb-vllm-tp1-sweep

qwen3-6-35b-a3b-gptq-4bit-intel-arc-pro-b70-32gb-vllm-tp1-sweep

Record

Recipe
qwen3-6-35b-a3b-gptq-4bit-intel-arc-pro-b70-32gb-vllm-tp1
Measured
2026-08-08T07:05:56.666Z
Points
1

Measured speed

ConcurrencyContextPrefillDecodeTTFT msStatus
32,76812,400248.3observed

Remaining fields

Identity, launch, related records, and measured speed are shown above. This is the rest of the normalized record.

id
qwen3-6-35b-a3b-gptq-4bit-intel-arc-pro-b70-32gb-vllm-tp1-sweep
measured at
2026-08-08T07:05:56.666Z
recipe id
qwen3-6-35b-a3b-gptq-4bit-intel-arc-pro-b70-32gb-vllm-tp1
schema version
local-ai-registry/v1

metrics

inference engine version
vLLM 0.26.1.dev0 XPU, built from v0.26.0 / 568afb3a1
latest point at
2026-08-08T07:05:56.666Z
max context tokens
32,768
peak generation tps
248.3
peak prompt tps
12,400
point count
1

rows

concurrencycontext tokensdecode tok sdecode tok s per streamoutput tokenspeak vram gbprefill tok ssamples
Unknown32,768248.3Unknown32Unknown12,4001
Provenance & metadata (1)