Speed sweep

pg-019f3295-e39e-785d-aef9-2738ae7a9ae0-sweep

pg-019f3295-e39e-785d-aef9-2738ae7a9ae0-sweep

JSON API Back to speed sweep search

Record

Recipe: pg-mlx-community-qwen3-6-27b-8bit-8bit-apple-m3-ultra-96gb-80c-mlx-f71cc81bde
Measured: 2026-07-05T14:01:51.769Z
Points: 4

Measured speed

Concurrency	Context	Prefill	Decode	TTFT ms	Status
1	8,256	—	20.2	—	observed
1	32,832	—	18.7	—	observed
1	258,111	—	11.1	—	observed
1	32,768	—	—	96,841.4	observed

Remaining fields

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

id: pg-019f3295-e39e-785d-aef9-2738ae7a9ae0-sweep
measured at: 2026-07-05T14:01:51.769Z
recipe id: pg-mlx-community-qwen3-6-27b-8bit-8bit-apple-m3-ultra-96gb-80c-mlx-f71cc81bde
schema version: local-ai-registry/v1

metrics

base memory bytes: 29218889827
base memory context tokens: 192
concurrency: 1
decode32k context tokens: 32,832
decode32k tps: 18.699
decode8k context tokens: 8,256
decode8k tps: 20.174
decode max context tokens: 258,111
decode max context tps: 11.115
decode mode: non-mtp
inference engine version: 0.31.3
latest point at: 2026-07-05T14:01:51.769Z
max context tokens: 258,111
max prompt tokens: 258,047
memory8k bytes: 32988980478
memory8k context tokens: 8,256
memory max context bytes: 89852486918
memory max context tokens: 258,111
peak generation tps: 18.699
peak memory bytes: 89852486918
peak prompt tps: 338.368
point count: 223
ttft32k cached prompt tokens: 28,736
ttft32k context tokens: 32,768
ttft32k seconds: 96.841

rows

concurrency	context tokens	decode tok s	decode tok s per stream	output tokens	peak vram gb	prefill tok s	samples
1	8,256	20.174	20.174	Unknown	30.723	Unknown	1
1	32,832	18.699	18.699	Unknown	83.682	Unknown	1
1	258,111	11.115	11.115	Unknown	83.682	Unknown	1
1	32,768	Unknown	Unknown	Unknown	Unknown	Unknown	1

Provenance & metadata (1)

source

paths: publication:pg-20260827T060320709Z, run:019f3295-e39e-785d-aef9-2738ae7a9ae0
repository: local.ai ↗