Compare runs
Pick two or more runs to compare. Qwen3.8-Flash-Next on M4 Max 128 GB is the first; add another below.
| Run | Qwen3.8-Flash-NextM4 Max 128 GB · llama.cppTake out Qwen3.8-Flash-Next |
|---|---|
| Score | 2,761Silver |
| Tasks passed | 44 of 44 |
| Points | 64 of 64 |
| Speed | 27.6 tok/s |
| Tokens | 127,182 |
| Summarizing | 5/5 |
| One-sentence summary | Passed21.5 tok/s |
| Meeting notes in three bullets | Passed24.9 tok/s |
| Review sentiment and complaint | Passed32.2 tok/s |
| Faithful summary of a study | Passed21.2 tok/s |
| Action items from an email thread | Passed29.5 tok/s |
| Extracting | 5/5 |
| Fields from an email as JSON | Passed26.4 tok/s |
| Every email address, once | Passed31.9 tok/s |
| A list as CSV | Passed29.9 tok/s |
| The real deadline | Passed27.1 tok/s |
| Warnings and errors from a log | Passed26.7 tok/s |
| Following instructions | 5/5 |
| Exactly one word | Passed23.6 tok/s |
| Three lowercase lines | Passed32.9 tok/s |
| Exact JSON shape | Passed28.2 tok/s |
| Sort a list | Passed30.3 tok/s |
| A name with three rules | Passed33 tok/s |
| Reasoning | 5/5 |
| Unit price | Passed29.5 tok/s |
| Days of the week | Passed26.8 tok/s |
| Who owns the cat | Passed32 tok/s |
| Arrival time | Passed18.9 tok/s |
| Filling a tank | Passed26.6 tok/s |
| Reading code | 4/4 |
| What does it print | Passed21.2 tok/s |
| Trace a loop | Passed22 tok/s |
| Find the bug | Passed26.2 tok/s |
| Run a query in your head | Passed23.4 tok/s |
| Using tools | 5/5 |
| Call one tool | Passed33.8 tok/s |
| Don't call a tool when none is needed | Passed18.4 tok/s |
| Two calls for two answers | Passed26.3 tok/s |
| Chain two tools | Passed28.4 tok/s |
| Handle a failed call honestly | Passed24.9 tok/s |
| Building pages | 5/5 |
| A small website | Passed27.7 tok/s |
| A pricing table with a toggle | Passed27 tok/s |
| An accessible sign-up form | Passed27.3 tok/s |
| A to-do app that remembers | Passed27.6 tok/s |
| A bar chart in SVG | Passed27.4 tok/s |
| Building games | 2/2 |
| Build 2048 | Passed27.6 tok/s |
| Build Snake | Passed27.1 tok/s |
| Writing code | 3/3 |
| Merge intervals | Passed34 tok/s |
| An LRU cache | Passed29.4 tok/s |
| Evaluate arithmetic | Passed29.4 tok/s |
| Fixing code | 2/2 |
| Fix the bookings module | Passed30.7 tok/s |
| Fix the CSV reader | Passed30.7 tok/s |
| Planning and data | 3/3 |
| A query over hidden data | Passed33.9 tok/s |
| Schedule five meetings | Passed28.6 tok/s |
| A replacement in ten calls or fewer | Passed24.2 tok/s |
Scrolls sideways to show every run.
A check marks the best of each row. tok/s is tokens a second: how fast the model writes. How the score works.