Speed sweep

pg-019f2c34-e1ba-7e39-b2d3-73d196f360d4-sweep

pg-019f2c34-e1ba-7e39-b2d3-73d196f360d4-sweep

JSON API Back to speed sweep search

Record

Recipe: pg-mlx-community-gemma-4-31b-it-4bit-4bit-apple-m5-max-128gb-mlx-eabfb61083
Measured: 2026-07-04T08:18:10.998Z
Points: 4

Measured speed

Concurrency	Context	Prefill	Decode	TTFT ms	Status
1	8,256	—	24.2	—	observed
1	32,832	—	19.4	—	observed
1	258,111	—	6.9	—	observed
1	32,768	—	—	67,344.7	observed

Remaining fields

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

id: pg-019f2c34-e1ba-7e39-b2d3-73d196f360d4-sweep
measured at: 2026-07-04T08:18:10.998Z
recipe id: pg-mlx-community-gemma-4-31b-it-4bit-4bit-apple-m5-max-128gb-mlx-eabfb61083
schema version: local-ai-registry/v1

metrics

base memory bytes: 17874145294
base memory context tokens: 192
concurrency: 1
decode32k context tokens: 32,832
decode32k tps: 19.449
decode8k context tokens: 8,256
decode8k tps: 24.227
decode max context tokens: 258,111
decode max context tps: 6.94
decode mode: non-mtp
inference engine version: 0.31.3
latest point at: 2026-07-04T08:18:10.998Z
max context tokens: 258,111
max prompt tokens: 258,047
memory8k bytes: 23321488630
memory8k context tokens: 8,256
memory max context bytes: 97041607926
memory max context tokens: 258,111
peak generation tps: 19.449
peak memory bytes: 97041607926
peak prompt tps: 486.571
point count: 223
ttft32k cached prompt tokens: 28,736
ttft32k context tokens: 32,768
ttft32k seconds: 67.345

rows

concurrency	context tokens	decode tok s	decode tok s per stream	output tokens	peak vram gb	prefill tok s	samples
1	8,256	24.227	24.227	Unknown	21.72	Unknown	1
1	32,832	19.449	19.449	Unknown	90.377	Unknown	1
1	258,111	6.94	6.94	Unknown	90.377	Unknown	1
1	32,768	Unknown	Unknown	Unknown	Unknown	Unknown	1

Provenance & metadata (1)

source

paths: publication:pg-20260827T060320709Z, run:019f2c34-e1ba-7e39-b2d3-73d196f360d4
repository: local.ai ↗