qwen3.8
ollama pull qwen3.8:<size> · sizes supported: latest
Every size listed here runs all six agent surfaces, five consecutive times each: 30 of 30 runs. Among the slowest on this list, and it reaches the same verdicts as models six times faster.
What it does, and how long it takes
Section titled “What it does, and how long it takes”Seconds are the median of the runs that passed, per surface.
| Size | Disk | Investigate | Optimize | Assess | Operate | Analyze | Plan | Median | Slowest |
|---|---|---|---|---|---|---|---|---|---|
| qwen3.8:latest | 19 GB | 31s | 2:19 | 2:50 | 1:12 | 1:38 | 57s | 1:12 | 3:15 |
Every cell is 5/5, so the table says how long rather than whether.
What it needs that the defaults do not give it
Section titled “What it needs that the defaults do not give it”qwen3.8:latest
Section titled “qwen3.8:latest”Nothing. Measured at the defaults — temperature 0, top_p 1, a ceiling of twelve unreported calls — and it clears every surface on them.
Method, and where these numbers stop being safe to generalise from:
methodology.md.