llama-3-70b-instruct
45±0
44interval45
aiqbench · LLM benchmarks
Benchmarks
- Sources
- 8
- Entries tracked
- 22,310
- Last sync
- 20 min20 minutes ago
All sources syncing normally
- Arena · Text· 8 hours ago
- Arena · WebDev· 8 hours ago
- Arena · Vision· 8 hours ago
- Arena · Agent· 8 hours ago
- GPQA Diamond· 20 minutes ago
- SWE-bench· 20 minutes ago
- Terminal-Bench· 20 minutes ago
- ARC-AGI-2· 20 minutes ago
Filter by org:
Released:
Meta-Llama-3-70B-Instruct
35±1
33interval36
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 | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 594 | llama-3-70b-instruct | meta | Apr 2024 | — | 44.6 28.3Ranked 44.6 = (1 × 28.3 + 3 × 50) ÷ 4. 28.3 is the mean percentile over the 1 of 22 selected axes this model is measured on; fewer axes means a stronger pull toward 50. | 44–45 | — | 1/22low | #284–594 | 28 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | ||
| 946 | Meta-Llama-3-70B-Instruct | Meta AI | — | — | 34.8 19.6Ranked 34.8 = (3 × 19.6 + 3 × 50) ÷ 6. 19.6 is the mean percentile over the 3 of 22 selected axes this model is measured on; fewer axes means a stronger pull toward 50. | 33–36 | 9 | 3/22 | #372–946 | — | — | — | — | — | — | — | 21 | 14 | — | — | — | — | — | 23 | — | — | — | — | — | — | — |
Blend is coverage-shrunk; the raw composite is shown greyed beside it · A low badge means fewer than 3 boards scored this model