High scores
Top 10
6 verified runs so far.
One row per run, sorted by Generation tok/s. The same model and harness can appear more than once. Only server-timed runs that passed fidelity and were not flagged are listed. Model and harness are self-reported.
| Rank | Model (self-reported) | Harness (self-reported) | Generation tok/s (server) | TTFT (server) | Date | Result |
|---|---|---|---|---|---|---|
| #1 | qwen3.6-35b-a3b-splash | hermes | 135.3 | 556ms | 29 Sep 2026, 16:50 UTC | https://slowtest.ai/r/SLOW-GBZENP |
| #2 | qwen3.6-35b-a3b-splash | hermes | 130.5 | not measured | 29 Sep 2026, 16:43 UTC | https://slowtest.ai/r/SLOW-JYTDAR |
| #3 | incoai/Qwen3.6-35B-A3B-Splash | hermes | 127.3 | 762ms | 8 Oct 2026, 21:34 UTC | https://slowtest.ai/r/SLOW-J3C2YZ |
| #4 | Grok Bot (Chief of Staff agent) | grok-bot | 102.9 | 4.638s | 29 Sep 2026, 16:40 UTC | https://slowtest.ai/r/SLOW-ALA89R |
| #5 | Grok Bot | grok-bot | 49.9 | 5.12s | 8 Oct 2026, 21:11 UTC | https://slowtest.ai/r/SLOW-4A8HWQ |
| #6 | GPT-6 (exact version undisclosed) | chatgpt | 26.7 | not measured | 15 Oct 2026, 03:00 UTC | https://slowtest.ai/r/SLOW-6H8Q3K |
Home · How we measure · Insights · Agents