Speed sweep

pg-019f3295-e48c-7585-9ae4-aa817e96e51b-sweep

pg-019f3295-e48c-7585-9ae4-aa817e96e51b-sweep

JSON API Back to speed sweep search

Record

Recipe: pg-mlx-community-gemma-4-31b-it-8bit-8bit-apple-m3-ultra-96gb-80c-mlx-ebbb61137d
Measured: 2026-07-05T14:01:52.006Z
Points: 4

Measured speed

Concurrency	Context	Prefill	Decode	TTFT ms	Status
1	8,256	—	17.2	—	observed
1	32,832	—	15.2	—	observed
1	258,111	—	7.1	—	observed
1	32,768	—	—	129,800.2	observed

Remaining fields

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

id: pg-019f3295-e48c-7585-9ae4-aa817e96e51b-sweep
measured at: 2026-07-05T14:01:52.006Z
recipe id: pg-mlx-community-gemma-4-31b-it-8bit-8bit-apple-m3-ultra-96gb-80c-mlx-ebbb61137d
schema version: local-ai-registry/v1

metrics

base memory bytes: 33155977196
base memory context tokens: 192
concurrency: 1
decode32k context tokens: 32,832
decode32k tps: 15.224
decode8k context tokens: 8,256
decode8k tps: 17.193
decode max context tokens: 258,111
decode max context tps: 7.092
decode mode: non-mtp
inference engine version: 0.31.3
latest point at: 2026-07-05T14:01:52.006Z
max context tokens: 258,111
max prompt tokens: 258,047
memory8k bytes: 38669495542
memory8k context tokens: 8,256
memory max context bytes: 112516558070
memory max context tokens: 258,111
peak generation tps: 15.224
peak memory bytes: 112516558070
peak prompt tps: 252.45
point count: 223
ttft32k cached prompt tokens: 28,736
ttft32k context tokens: 32,768
ttft32k seconds: 129.8

rows

concurrency	context tokens	decode tok s	decode tok s per stream	output tokens	peak vram gb	prefill tok s	samples
1	8,256	17.193	17.193	Unknown	36.014	Unknown	1
1	32,832	15.224	15.224	Unknown	104.789	Unknown	1
1	258,111	7.092	7.092	Unknown	104.789	Unknown	1
1	32,768	Unknown	Unknown	Unknown	Unknown	Unknown	1

Provenance & metadata (1)

source

paths: publication:pg-20260827T060320709Z, run:019f3295-e48c-7585-9ae4-aa817e96e51b
repository: local.ai ↗