Speed sweep

pg-019f389a-be34-7874-a34a-3ec3e97a8d4b-sweep

pg-019f389a-be34-7874-a34a-3ec3e97a8d4b-sweep

JSON API Back to speed sweep search

Record

Recipe: pg-mlx-community-gemma-4-26b-a4b-it-8bit-8bit-apple-m5-pro-64gb-mlx-ead128b2bb
Measured: 2026-07-06T18:04:53.041Z
Points: 4

Measured speed

Concurrency	Context	Prefill	Decode	TTFT ms	Status
1	8,256	—	46.6	—	observed
1	32,832	—	37.7	—	observed
1	90,432	—	26.2	—	observed
1	32,768	—	—	24,222.5	observed

Remaining fields

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

id: pg-019f389a-be34-7874-a34a-3ec3e97a8d4b-sweep
measured at: 2026-07-06T18:04:53.041Z
recipe id: pg-mlx-community-gemma-4-26b-a4b-it-8bit-8bit-apple-m5-pro-64gb-mlx-ead128b2bb
schema version: local-ai-registry/v1

metrics

base memory bytes: 27017822504
base memory context tokens: 192
concurrency: 1
decode32k context tokens: 32,832
decode32k tps: 37.734
decode8k context tokens: 8,256
decode8k tps: 46.641
decode max context tokens: 90,432
decode max context tps: 26.245
decode mode: non-mtp
inference engine version: unknown
latest point at: 2026-07-06T18:04:53.041Z
max context tokens: 90,432
max prompt tokens: 90,368
memory8k bytes: 29221340462
memory8k context tokens: 8,256
memory max context bytes: 38014821896
memory max context tokens: 90,432
peak generation tps: 37.734
peak memory bytes: 38014821896
peak prompt tps: 1,352.792
point count: 79
ttft32k cached prompt tokens: 28,736
ttft32k context tokens: 32,768
ttft32k seconds: 24.222

rows

concurrency	context tokens	decode tok s	decode tok s per stream	output tokens	peak vram gb	prefill tok s	samples
1	8,256	46.641	46.641	Unknown	27.214	Unknown	1
1	32,832	37.734	37.734	Unknown	35.404	Unknown	1
1	90,432	26.245	26.245	Unknown	35.404	Unknown	1
1	32,768	Unknown	Unknown	Unknown	Unknown	Unknown	1

Provenance & metadata (1)

source

paths: publication:pg-20260827T060320709Z, run:019f389a-be34-7874-a34a-3ec3e97a8d4b
repository: local.ai ↗