Speed sweep

pg-019f34f2-f8f0-7e14-8756-62f38dfc6b51-sweep

pg-019f34f2-f8f0-7e14-8756-62f38dfc6b51-sweep

JSON API Back to speed sweep search

Record

Recipe: pg-mlx-community-gemma-4-26b-a4b-it-4bit-4bit-apple-m3-ultra-96gb-60c-mlx-ef6ec29fa2
Measured: 2026-07-06T01:02:46.044Z
Points: 4

Measured speed

Concurrency	Context	Prefill	Decode	TTFT ms	Status
1	8,256	—	83.6	—	observed
1	32,832	—	66.3	—	observed
1	262,079	—	20.6	—	observed
1	32,768	—	—	26,977.5	observed

Remaining fields

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

id: pg-019f34f2-f8f0-7e14-8756-62f38dfc6b51-sweep
measured at: 2026-07-06T01:02:46.044Z
recipe id: pg-mlx-community-gemma-4-26b-a4b-it-4bit-4bit-apple-m3-ultra-96gb-60c-mlx-ef6ec29fa2
schema version: local-ai-registry/v1

metrics

base memory bytes: 14473936288
base memory context tokens: 192
concurrency: 1
decode32k context tokens: 32,832
decode32k tps: 66.279
decode8k context tokens: 8,256
decode8k tps: 83.601
decode max context tokens: 262,079
decode max context tps: 20.624
decode mode: non-mtp
inference engine version: 0.31.3
latest point at: 2026-07-06T01:02:46.044Z
max context tokens: 262,079
max prompt tokens: 262,015
memory8k bytes: 16611261490
memory8k context tokens: 8,256
memory max context bytes: 43412989564
memory max context tokens: 262,079
peak generation tps: 66.279
peak memory bytes: 43412989564
peak prompt tps: 1,214.644
point count: 224
ttft32k cached prompt tokens: 28,736
ttft32k context tokens: 32,768
ttft32k seconds: 26.977

rows

concurrency	context tokens	decode tok s	decode tok s per stream	output tokens	peak vram gb	prefill tok s	samples
1	8,256	83.601	83.601	Unknown	15.47	Unknown	1
1	32,832	66.279	66.279	Unknown	40.431	Unknown	1
1	262,079	20.624	20.624	Unknown	40.431	Unknown	1
1	32,768	Unknown	Unknown	Unknown	Unknown	Unknown	1

Provenance & metadata (1)

source

paths: publication:pg-20260827T060320709Z, run:019f34f2-f8f0-7e14-8756-62f38dfc6b51
repository: local.ai ↗