← All runs

Evaluation run

self-correction chains

6 July 2026, 13:53 · eval-engine · 232478e · 89 cases

Did it hear the words right?

This run graded typed text only, so there is nothing here to have misheard.

Did it clean up correctly and safely?

96%
Light, ships on by default
85/89 cases passed · 231 ms typical
92%
Polish, experimental
82/89 cases passed · 265 ms typical
When a case fails, whose fault is it?
0 the microphone misheard the words  ·  11 the cleanup itself slipped

Every case

Each case shows the instant on-device cleanup, then what each mode did to it. Added words are marked, removed words are struck. For a spoken case, Heard is what the microphone transcribed and Was said is the reference it is scored against.

89 cases match

num-timeTypedclean
Instantlet's meet at 4:30 this afternoon
LightLet's meet at 4:30 this afternoon. 261 ms
PolishLet's meet at 4:30 this afternoon. 293 ms
num-time-amTypedclean
Instantthe meeting is at 9 am tomorrow
LightThe meeting is at 9 am tomorrow. 231 ms
PolishThe meeting is at 9 AM tomorrow. 243 ms
num-versionTypedclean
Instantplease update to version 2 point one
LightPlease update to version 2 point one 2.1. 272 ms
PolishPlease update to version 2 point one 2.1. 215 ms
num-version-32Typedclean
Instantupgrade to version 3 point 2 please
LightUpgrade to version 3 point 2 3.2 please. 228 ms
PolishPlease upgrade to version 3 point 2 please 3.2. 244 ms
num-yearTypedclean
Instantwe launched back in twenty twenty
LightWe launched back in twenty twenty 2020. 224 ms
PolishWe launched back in twenty twenty 2020. 227 ms
num-year-2024Typedclean
Instantwe started this back in twenty twenty four
LightWe started this back in twenty twenty four 2024. 258 ms
PolishWe started this back in twenty twenty four 2024. 335 ms
punct-questionTypedclean
Instantare we still on for lunch today
LightAre we still on for lunch today? 201 ms
PolishAre we still on for lunch today? 252 ms
punct-runonTypedclean
Instantthe build passed we can ship it tell the team
LightThe build passed, we can ship it, tell the team. 325 ms
PolishThe build passed. We can ship it and tell the team. 336 ms
punct-runon-2Typedclean
Instanti went to the store i bought milk i came back home
LightI went to the store i and bought milk; I came back home. 389 ms
PolishI went to the store i and bought milk. Then I came back home. 355 ms
real-emailTypedclean
Instanthey sarah just wanted to follow up on the proposal do you have time to hop on a quick call tomorrow around 3
LightHey Sarah, just wanted to follow up on the proposal. Do you have time to hop on a quick call tomorrow around 3? 549 ms
PolishHey Sarah, just wanted to follow up on the proposal. Do you have time to hop on for a quick call tomorrow around 3? 533 ms