Speed sweep

lfm2-8b-a1b-q4-k-m-rtx-3060-12gb-llama-cpp-tp2-geyuhhjh-sweep

lfm2-8b-a1b-q4-k-m-rtx-3060-12gb-llama-cpp-tp2-geyuhhjh-sweep

Record

Recipe
lfm2-8b-a1b-q4-k-m-rtx-3060-12gb-llama-cpp-tp2-geyuhhjh
Measured
2026-05-28T22:13:31.516Z
Points
1

Measured speed

ConcurrencyContextPrefillDecodeTTFT msStatus
1128,0006,948.3193.5observed

Remaining fields

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

id
lfm2-8b-a1b-q4-k-m-rtx-3060-12gb-llama-cpp-tp2-geyuhhjh-sweep
measured at
2026-05-28T22:13:31.516Z
recipe id
lfm2-8b-a1b-q4-k-m-rtx-3060-12gb-llama-cpp-tp2-geyuhhjh
schema version
local-ai-registry/v1

metrics

concurrency
1
latest point at
2026-05-28T22:13:31.516Z
max context tokens
128,000
peak generation tps
193.5
peak prompt tps
6,948.3
point count
1

rows

concurrencycontext tokensdecode tok sdecode tok s per streamoutput tokenspeak vram gbprefill tok ssamples
1128,000193.5Unknown2567.66,948.31
Provenance & metadata (1)