# LLM Leaderboard History

> Server-readable mirror of https://benchlm.ai/llm-leaderboard-history using Arena Leaderboard Dataset snapshots fetched on 2026-09-02.

## Key Numbers

- First tracked frontier: vicuna-13b at 1075 Elo in 2023-05
- Current frontier: claude-opus-5-max at 1505 Elo in 2026-09
- Total Elo gain: +430
- Crown changes since 2023-05: 19
- Provider with most months at #1: openai (15 months)

## Current Top Arena Models (2026-09)

| Rank | Model | Org | License | Elo | Votes |
|------|-------|-----|---------|-----|-------|
| 1 | claude-opus-5-max | anthropic | Proprietary | 1505 | 16839 |
| 2 | claude-opus-5-high | anthropic | Proprietary | 1503.6 | 34617 |
| 3 | claude-opus-4-6-high | anthropic | Proprietary | 1503 | 72110 |
| 4 | claude-opus-4-6 | anthropic | Proprietary | 1497.7 | 76032 |
| 5 | claude-fable-5 | anthropic | Proprietary | 1494.4 | 26977 |
| 6 | gemini-3.7-flash-high | google | Proprietary | 1491.1 | 5685 |
| 7 | claude-opus-4-7-high | anthropic | Proprietary | 1490.1 | 60123 |
| 8 | muse-spark-1.2 (xHigh) | meta | Proprietary | 1488.5 | 3244 |
| 9 | gemini-3.5-flash-high | google | Proprietary | 1483.5 | 33874 |
| 10 | claude-opus-4-7 | anthropic | Proprietary | 1483.2 | 61256 |
| 11 | gemini-3.1-pro-preview | google | Proprietary | 1480.1 | 102763 |
| 12 | qwen3.8-max | alibaba | Proprietary | 1479.6 | 13136 |
| 13 | gemini-3-pro | google | Proprietary | 1479.5 | 40675 |
| 14 | muse-spark-1.1 | meta | Proprietary | 1479 | 23768 |
| 15 | gemini-3.6-flash-high | google | Proprietary | 1476.4 | 22009 |
| 16 | kimi-k3-max | moonshot | Kimi K3 license | 1476.3 | 17895 |
| 17 | gemini-3.5-flash-medium | google | Proprietary | 1475.7 | 32274 |
| 18 | glm-5.3-max | zai | MIT | 1474.5 | 7401 |
| 19 | qwen3.7-max-preview | alibaba | Proprietary | 1474.3 | 3710 |
| 20 | muse-spark | meta | Proprietary | 1473.6 | 13571 |

## Recent Open vs Proprietary Gap

| Month | Proprietary #1 | Open #1 | Gap |
|-------|----------------|---------|-----|
| 2025-10 | gemini-2.5-pro (1464.7) | glm-4.6 (1438.5) | 26.2 |
| 2025-11 | gemini-3-pro (1487.6) | glm-4.6 (1441.8) | 45.8 |
| 2025-12 | gemini-3-pro (1484.4) | glm-4.6 (1439.9) | 44.5 |
| 2026-01 | gemini-3-pro (1481.2) | kimi-k2.5-thinking (1445.9) | 35.3 |
| 2026-02 | claude-opus-4-6 (1501.6) | qwen3.5-397b-a17b (1451.6) | 50 |
| 2026-03 | claude-opus-4-6-thinking (1500.6) | kimi-k2.5-thinking (1449.5) | 51.1 |
| 2026-04 | claude-opus-4-6-thinking (1500.6) | glm-5.1 (1467.7) | 32.9 |
| 2026-05 | claude-opus-4-6-thinking (1499.3) | glm-5.1 (1469.3) | 30 |
| 2026-06 | claude-opus-4-6-thinking (1500.3) | glm-5.1 (1468.3) | 32 |
| 2026-07 | claude-opus-5-max (1511.6) | kimi-k3-max (1473.3) | 38.3 |
| 2026-08 | claude-opus-5-max (1504.7) | kimi-k3-max (1476.1) | 28.6 |
| 2026-09 | claude-opus-5-max (1505) | kimi-k3-max (1476.3) | 28.7 |

## Recent Crown Changes

| Month | From | To | Winning Org |
|-------|------|----|-------------|
| 2026-07 | claude-opus-4-6-thinking | claude-opus-5-max | anthropic |
| 2026-03 | claude-opus-4-6 | claude-opus-4-6-thinking | anthropic |
| 2026-02 | gemini-3-pro | claude-opus-4-6 | anthropic |
| 2025-11 | gemini-2.5-pro | gemini-3-pro | google |
| 2025-06 | o3-2025-04-16 | gemini-2.5-pro | google |
| 2025-04 | chatgpt-4o-latest-20250326 | o3-2025-04-16 | openai |
| 2025-03 | deepseek-r1 | chatgpt-4o-latest-20250326 | openai |
| 2025-01 | o1-2024-12-17 | deepseek-r1 | deepseek |
| 2024-12 | o1-preview | o1-2024-12-17 | openai |
| 2024-09 | chatgpt-4o-latest | o1-preview | openai |
| 2024-08 | gpt-4o-2024-05-13 | chatgpt-4o-latest | openai |
| 2024-05 | gpt-4-turbo-2024-04-09 | gpt-4o-2024-05-13 | openai |
| 2024-04 | claude-3-opus-20240229 | gpt-4-turbo-2024-04-09 | openai |
| 2024-03 | gpt-4-1106-preview | claude-3-opus-20240229 | anthropic |
| 2024-02 | gpt-4-0314 | gpt-4-1106-preview | openai |
| 2023-12 | wizardlm-70b | gpt-4-0314 | openai |
| 2023-10 | vicuna-33b | wizardlm-70b | microsoft |
| 2023-07 | guanaco-33b | vicuna-33b | LMSYS |
| 2023-06 | vicuna-13b | guanaco-33b | UW |

Canonical page: https://benchlm.ai/llm-leaderboard-history