Commit ·
7bfb750
1
Parent(s): a0493f2
docs: add multimodal Multi-O Eval leaderboard
Browse files
README.md
CHANGED
|
@@ -81,6 +81,18 @@ The released board in the GitHub repository additionally reports a **Multi-O Eva
|
|
| 81 |
averages each task over every O-award human paper that reports the task's scored field, and
|
| 82 |
a **Multi-BO index** anchored at DeepSeek V4.1 Flash = 1.000.
|
| 83 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 84 |
## Quickstart
|
| 85 |
|
| 86 |
```bash
|
|
|
|
| 81 |
averages each task over every O-award human paper that reports the task's scored field, and
|
| 82 |
a **Multi-BO index** anchored at DeepSeek V4.1 Flash = 1.000.
|
| 83 |
|
| 84 |
+
## Leaderboard
|
| 85 |
+
|
| 86 |
+
| Rank | Model | Multi-O Eval | Multi-BO index (Flash = 1.000) |
|
| 87 |
+
|---:|---|---:|---:|
|
| 88 |
+
| 1 | claude-code + Fable 5.1 | **64.62%** | 1.1637 |
|
| 89 |
+
| 2 | codex + GPT-6 Astra | **61.47%** | 1.1200 |
|
| 90 |
+
| 3 | opencode + DeepSeek V4.1 Flash | **59.81%** | 1.0000 |
|
| 91 |
+
| 4 | codex + GPT-5.5 xhigh | **59.75%** | 1.0061 |
|
| 92 |
+
| 5 | opencode + DeepSeek V4.1 Flash (max) | **48.11%** | 0.8636 |
|
| 93 |
+
|
| 94 |
+
Single runs; across 3 seeds the per-model deviation is ±1.3–2.9pp.
|
| 95 |
+
|
| 96 |
## Quickstart
|
| 97 |
|
| 98 |
```bash
|