hermes-3-llama-3.1-70b
37±0
37interval38
aiqbench · LLM benchmarks
Benchmarks
- Sources
- 8
- Entries tracked
- 22,310
- Last sync
- 35 min35 minutes ago
All sources syncing normally
- Arena · Text· 1 hour ago
- Arena · WebDev· 1 hour ago
- Arena · Vision· 1 hour ago
- Arena · Agent· 1 hour ago
- GPQA Diamond· 35 minutes ago
- SWE-bench· 35 minutes ago
- Terminal-Bench· 35 minutes ago
- ARC-AGI-2· 35 minutes ago
Filter by org:
Released:
One grid, every column. Pick axes in the Axes panel — one axis ranks like a single board, several blend into a weighted composite, all of them read as the index. Rank is assigned before any filter and never renumbers.
| Model | Org | Interval | Swing | i | i | i | i | i | i | i | i | i | i | i | i | i | i | i | i | i | i | i | i | i | i | Add to compare | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 908 | hermes-3-llama-3.1-70b | Nous Research | Jul 2024 | $0.70 / $0.70 | 37.2 24.5Ranked 37.2 = (3 × 24.5 + 3 × 50) ÷ 6. 24.5 is the mean percentile over the 3 of 22 selected axes this model is measured on; fewer axes means a stronger pull toward 50. | — | 3 | 3/22 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 26 | — | — | — | 23 | 24 | — | — |
Blend is coverage-shrunk; the raw composite is shown greyed beside it · A low badge means fewer than 3 boards scored this model