Speed sweep

pg-019f7582-f1df-712d-b08e-1c004f8d096e-sweep

pg-019f7582-f1df-712d-b08e-1c004f8d096e-sweep

JSON API Back to speed sweep search

Record

Recipe: pg-mlx-community-agents-a1-8bit-8bit-apple-m3-ultra-96gb-60c-mlx-ff5a376706
Measured: 2026-07-18T13:55:43.682Z
Points: 4

Measured speed

Concurrency	Context	Prefill	Decode	TTFT ms	Status
1	8,256	—	71.4	—	observed
1	32,832	—	61.2	—	observed
1	65,600	—	54.1	—	observed
1	32,768	—	—	24,852.5	observed

Remaining fields

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

id: pg-019f7582-f1df-712d-b08e-1c004f8d096e-sweep
measured at: 2026-07-18T13:55:43.682Z
recipe id: pg-mlx-community-agents-a1-8bit-8bit-apple-m3-ultra-96gb-60c-mlx-ff5a376706
schema version: local-ai-registry/v1

metrics

base memory bytes: 37134480110
base memory context tokens: 192
concurrency: 1
decode32k context tokens: 32,832
decode32k tps: 61.204
decode8k context tokens: 8,256
decode8k tps: 71.353
decode max context tokens: 65,600
decode max context tps: 54.091
decode mode: non-mtp
inference engine version: 0.31.3
latest point at: 2026-07-18T13:55:43.682Z
max context tokens: 65,600
max prompt tokens: 65,536
memory8k bytes: 39294965790
memory8k context tokens: 8,256
memory max context bytes: 45355170082
memory max context tokens: 65,600
peak generation tps: 61.204
peak memory bytes: 45355170082
peak prompt tps: 1,318.498
point count: 56
ttft32k cached prompt tokens: 28,736
ttft32k context tokens: 32,768
ttft32k seconds: 24.853

rows

concurrency	context tokens	decode tok s	decode tok s per stream	output tokens	peak vram gb	prefill tok s	samples
1	8,256	71.353	71.353	Unknown	36.596	Unknown	1
1	32,832	61.204	61.204	Unknown	42.24	Unknown	1
1	65,600	54.091	54.091	Unknown	42.24	Unknown	1
1	32,768	Unknown	Unknown	Unknown	Unknown	Unknown	1

Provenance & metadata (1)

source

paths: publication:pg-20260827T060320709Z, run:019f7582-f1df-712d-b08e-1c004f8d096e
repository: local.ai ↗