Speed sweep

pg-019f3295-e277-7fc6-99c3-e5247d50d1ee-sweep

pg-019f3295-e277-7fc6-99c3-e5247d50d1ee-sweep

JSON API Back to speed sweep search

Record

Recipe: pg-mlx-community-nemotron-3-ultra-550b-a55b-unknown-apple-m3-ultra-96gb-80c-mlx-3581a2e60f
Measured: 2026-07-05T14:01:51.474Z
Points: 4

Measured speed

Concurrency	Context	Prefill	Decode	TTFT ms	Status
1	8,256	—	99	—	observed
1	32,832	—	81	—	observed
1	127,040	—	47.7	—	observed
1	32,768	—	—	124,839.8	observed

Remaining fields

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

id: pg-019f3295-e277-7fc6-99c3-e5247d50d1ee-sweep
measured at: 2026-07-05T14:01:51.474Z
recipe id: pg-mlx-community-nemotron-3-ultra-550b-a55b-unknown-apple-m3-ultra-96gb-80c-mlx-3581a2e60f
schema version: local-ai-registry/v1

metrics

base memory bytes: 12237920668
base memory context tokens: 192
concurrency: 1
decode32k context tokens: 32,832
decode32k tps: 80.987
decode8k context tokens: 8,256
decode8k tps: 99.006
decode max context tokens: 127,040
decode max context tps: 47.717
decode mode: non-mtp
inference engine version: unknown
latest point at: 2026-07-05T14:01:51.474Z
max context tokens: 127,040
max prompt tokens: 126,976
memory8k bytes: 13256377814
memory8k context tokens: 8,256
memory max context bytes: 19143525334
memory max context tokens: 127,040
peak generation tps: 70.528
peak memory bytes: 19143525334
peak prompt tps: 36.212
point count: 111
ttft32k cached prompt tokens: 0
ttft32k context tokens: 32,768
ttft32k seconds: 124.84

rows

concurrency	context tokens	decode tok s	decode tok s per stream	output tokens	peak vram gb	prefill tok s	samples
1	8,256	99.006	99.006	Unknown	12.346	Unknown	1
1	32,832	80.987	80.987	Unknown	17.829	Unknown	1
1	127,040	47.717	47.717	Unknown	17.829	Unknown	1
1	32,768	Unknown	Unknown	Unknown	Unknown	Unknown	1

Provenance & metadata (1)

source

paths: publication:pg-20260827T060320709Z, run:019f3295-e277-7fc6-99c3-e5247d50d1ee
repository: local.ai ↗