Speed sweep

pg-019f3295-e322-7294-a6c2-f8d6285631ac-sweep

pg-019f3295-e322-7294-a6c2-f8d6285631ac-sweep

JSON API Back to speed sweep search

Record

Recipe: pg-mlx-community-qwen3-5-9b-optiq-4bit-4bit-apple-m3-ultra-96gb-80c-mlx-1b986c89c5
Measured: 2026-07-05T14:01:51.645Z
Points: 4

Measured speed

Concurrency	Context	Prefill	Decode	TTFT ms	Status
1	8,256	—	76.2	—	observed
1	32,832	—	67.7	—	observed
1	258,111	—	34.3	—	observed
1	32,768	—	—	28,829.7	observed

Remaining fields

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

id: pg-019f3295-e322-7294-a6c2-f8d6285631ac-sweep
measured at: 2026-07-05T14:01:51.645Z
recipe id: pg-mlx-community-qwen3-5-9b-optiq-4bit-4bit-apple-m3-ultra-96gb-80c-mlx-1b986c89c5
schema version: local-ai-registry/v1

metrics

base memory bytes: 7534259858
base memory context tokens: 192
concurrency: 1
decode32k context tokens: 32,832
decode32k tps: 67.729
decode8k context tokens: 8,256
decode8k tps: 76.226
decode max context tokens: 258,111
decode max context tps: 34.313
decode mode: non-mtp
inference engine version: unknown
latest point at: 2026-07-05T14:01:51.645Z
max context tokens: 258,111
max prompt tokens: 258,047
memory8k bytes: 9578154654
memory8k context tokens: 8,256
memory max context bytes: 42521971378
memory max context tokens: 258,111
peak generation tps: 67.729
peak memory bytes: 42521971378
peak prompt tps: 1,136.604
point count: 223
ttft32k cached prompt tokens: 28,736
ttft32k context tokens: 32,768
ttft32k seconds: 28.83

rows

concurrency	context tokens	decode tok s	decode tok s per stream	output tokens	peak vram gb	prefill tok s	samples
1	8,256	76.226	76.226	Unknown	8.92	Unknown	1
1	32,832	67.729	67.729	Unknown	39.602	Unknown	1
1	258,111	34.313	34.313	Unknown	39.602	Unknown	1
1	32,768	Unknown	Unknown	Unknown	Unknown	Unknown	1

Provenance & metadata (1)

source

paths: publication:pg-20260827T060320709Z, run:019f3295-e322-7294-a6c2-f8d6285631ac
repository: local.ai ↗