Compare runs
Pick two or more runs to compare. Llama3.2-Instruct on M3 Pro 18 GB is the first; add another below.
No other run of Llama3.2-Instruct or on M3 Pro 18 GB to add yet.
Scrolls sideways to show every run.
| RunA check marks the best of each row. | Llama3.2-InstructM3 Pro 18 GB · llama.cppRemove Llama3.2-Instruct |
|---|---|
| Score | 2,170Silver |
| Tasks passed | 15 of 24 |
| Speed | 34.7 tok/s |
| Tokens | 5,132 |
| Summarizing | 3/5 |
| One-sentence summary | Passed33.9 tok/s |
| Meeting notes in three bullets | Passed33.5 tok/s |
| Review sentiment and complaint | Failed35.5 tok/s |
| Faithful summary of a study | Passed33.5 tok/s |
| Action items from an email thread | Failed34 tok/s |
| Extracting | 2/5 |
| Fields from an email as JSON | Failed32.9 tok/s |
| Every email address, once | Passed35.3 tok/s |
| A list as CSV | Failed34.9 tok/s |
| The real deadline | Passed39.6 tok/s |
| Warnings and errors from a log | Failed34.7 tok/s |
| Following instructions | 5/5 |
| Exactly one word | Passed63.9 tok/s |
| Three lowercase lines | Passed35.1 tok/s |
| Exact JSON shape | Passed35.8 tok/s |
| Sort a list | Passed36.7 tok/s |
| A name with three rules | Passed40.4 tok/s |
| Reasoning | 3/5 |
| Unit price | Passed34.8 tok/s |
| Day and date | Failed35 tok/s |
| Who owns each pet | Failed34.8 tok/s |
| Arrival time | Passed35.1 tok/s |
| Filling a tank | Passed34.7 tok/s |
| Reading code | 2/4 |
| What does it print | Passed34.8 tok/s |
| Trace a loop | Failed34.6 tok/s |
| Find the bug | Passed35.2 tok/s |
| Run a query in your head | Failed36.7 tok/s |