eval-harness

model

Phi-4 14B llm

A model record: onboarded once for the project and used by any eval's arms.

weights
open
licence
MIT
aliases
phi4:14b
settings
defaults
onboarding
onboarded, no arm smoked 2026-10-31
best result per goal
#20 of 55 Triage decisions on Orrery's error signals

Every model

01deployments and their onboarding

x3 · x3 · ollama 0.40.13 · phi4:14b · onboarding pass 2026-10-31 · 22.4 tokens/s
where
x3
server or device
ollama 0.40.13
build
phi4:14b
ac896e5b8b34a1f4efa7b14d7520725140d5512484457fab45d2a4ea14c69dba
Q4_K_M
matches the record
yes
checked
2026-10-31
onboarded on
x3
probemeasurementflag
endpointprovider openai · base url http://127.0.0.1:11434/v1 · model phi4:14b—
digest foundac896e5b8b34a1f4efa7b14d7520725140d5512484457fab45d2a4ea14c69dba—
defaults 5.53 · tokens out 28 · reasoned: no · finish stop—
chosens 1.25 · tokens out 28 · reasoned: no · finish stop · settings ()—
tokens per s22.4—
deterministicyes—
json validyes—
dgx · dgx · ollama 0.40.13 · phi4:14b · onboarding pass 2026-10-31 · 23.0 tokens/s
where
dgx
server or device
ollama 0.40.13
build
phi4:14b
ac896e5b8b34a1f4efa7b14d7520725140d5512484457fab45d2a4ea14c69dba
matches the record
yes
checked
2026-10-31
onboarded on
dgx
probemeasurementflag
endpointprovider openai · base url http://100.69.49.126:11434/v1 · model phi4:14b—
digest foundac896e5b8b34a1f4efa7b14d7520725140d5512484457fab45d2a4ea14c69dba—
defaults 6.06 · tokens out 28 · reasoned: no · finish stop—
chosens 1.22 · tokens out 28 · reasoned: no · finish stop · settings ()—
tokens per s23.0—
deterministicyes—
json validyes—
mac · mac · ollama 0.40.13 · phi4:14b · onboarding pass 2026-10-31 · 21.9 tokens/s
where
mac
server or device
ollama 0.40.13
build
phi4:14b
ac896e5b8b34a1f4efa7b14d7520725140d5512484457fab45d2a4ea14c69dba
Q4_K_M
matches the record
yes
checked
2026-10-31
onboarded on
mac
probemeasurementflag
endpointprovider openai · base url http://127.0.0.1:11435/v1 · model phi4:14b—
digest foundac896e5b8b34a1f4efa7b14d7520725140d5512484457fab45d2a4ea14c69dba—
defaults 8.97 · tokens out 28 · reasoned: no · finish stop—
chosens 1.28 · tokens out 28 · reasoned: no · finish stop · settings ()—
tokens per s21.9—
deterministicyes—
json validyes—

What it can do, across its deployments: reasons: no · json mode: yes

02settings history

No change of settings or build is recorded.

03arms