Speed sweep

pg-01a011ee-c3ca-7798-997b-83dc65548a37-sweep

pg-01a011ee-c3ca-7798-997b-83dc65548a37-sweep

JSON API Back to speed sweep search

Record

Recipe: pg-antirez-deepseek-v4-gguf-iq2xxs-apple-m3-ultra-96gb-80c-llama-cpp-2a5a8c4a6c
Measured: 2026-08-17T22:54:15.493Z
Points: 3

Measured speed

Concurrency	Context	Prefill	Decode	TTFT ms	Status
1	8,256	—	22.5	—	observed
1	32,832	—	18.4	270,249	observed
1	221,504	—	8.3	—	observed

Remaining fields

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

id: pg-01a011ee-c3ca-7798-997b-83dc65548a37-sweep
measured at: 2026-08-17T22:54:15.493Z
recipe id: pg-antirez-deepseek-v4-gguf-iq2xxs-apple-m3-ultra-96gb-80c-llama-cpp-2a5a8c4a6c
schema version: local-ai-registry/v1

metrics

base memory bytes: 88753324032
base memory context tokens: 192
concurrency: 1
decode32k context tokens: 32,832
decode32k tps: 18.433
decode8k context tokens: 8,256
decode8k tps: 22.538
decode max context tokens: 221,504
decode max context tps: 8.287
decode mode: non-mtp
inference engine version: dsv4-fork (antirez DeepSeek-V4 fork)
latest point at: 2026-08-17T22:54:15.493Z
max context tokens: 221,504
max prompt tokens: 221,440
memory8k bytes: 93146152960
memory8k context tokens: 8,256
memory max context bytes: 154447986688
memory max context tokens: 221,504
peak generation tps: 22.538
peak memory bytes: 162733506560
peak prompt tps: 182.301
point count: 191
ttft32k cached prompt tokens: 28,735
ttft32k context tokens: 32,768
ttft32k seconds: 270.249

rows

concurrency	context tokens	decode tok s	decode tok s per stream	output tokens	peak vram gb	prefill tok s	samples
1	8,256	22.538	Unknown	Unknown	86.749	Unknown	1
1	32,832	18.433	Unknown	Unknown	151.557	Unknown	1
1	221,504	8.287	Unknown	Unknown	143.841	Unknown	1

Provenance & metadata (1)

source

paths: publication:pg-20260827T060320709Z, run:01a011ee-c3ca-7798-997b-83dc65548a37
repository: local.ai ↗