YZ Index
Weekly Report
Weekly model performance changes and trend analysis.
Baseline: Run #273 · Formula v7 · Judge v6.4 · Benchmark v7 · 2026-08-10 05:05 SGT
Current: Run #282 · Formula v7 · Judge v6.4 · Benchmark v7 · 2026-08-17 05:04 SGT
Overall Score Changes Ranked by absolute change magnitude
Claude Sonnet 4.6
+9.3
70.8 → 80.1
Gemini 2.5 Pro
+8.5
63.2 → 71.7
GLM-4.6
+8.5
56.1 → 64.5
Qwen3 Max
+5.7
67.6 → 73.3
GPT-5.5
+4.7
75.9 → 80.6
Grok 4
+0.9
79.2 → 80.1
GPT-o3
+0.9
78.9 → 79.8
DeepSeek V4 Pro
-8.0
78.8 → 70.8
Claude Opus 4.7
-1.7
87.5 → 85.8
1 — stable
Gemini 3.1 Pro (66.7)
Side Dimension Changes Communication and Judgment changes
Grok 4
+21.1
Judgment: 59.2 → 80.3
Grok 4
+9.8
Communication: 78.5 → 88.3
Claude Sonnet 4.6
+7.4
Communication: 82.5 → 89.9
GLM-4.6
+5.3
Judgment: 21.4 → 26.7
Gemini 3.1 Pro
+5.0
Judgment: 72.8 → 77.8
GPT-o3
+2.5
Communication: 78.3 → 80.8
DeepSeek V4 Pro
-25.0
Communication: 93.3 → 68.3
Claude Opus 4.7
-12.5
Communication: 85.3 → 72.8
Gemini 3.1 Pro
-12.5
Communication: 75.8 → 63.3
DeepSeek V4 Pro
-7.8
Judgment: 85.7 → 77.9
Gemini 2.5 Pro
-7.5
Communication: 75.8 → 68.3
Claude Sonnet 4.6
-6.1
Judgment: 83.7 → 77.6
Gemini 2.5 Pro
-6.0
Judgment: 72.8 → 66.8
Claude Opus 4.7
-2.8
Judgment: 86.1 → 83.3
Qwen3 Max
-2.5
Communication: 50.3 → 47.8
GPT-o3
-1.8
Judgment: 81.3 → 79.5
Operational Signal Changes Stability and Availability changes
Doubao Pro
+97.0
Availability: 0.0 → 97.0
Doubao Pro
+94.1
Value: 0.0 → 94.1
GLM-4.6
+11.0
Stability: 24.2 → 35.2
Grok 4
+10.6
Stability: 28.8 → 39.4
DeepSeek V4 Pro
+10.5
Stability: 37.4 → 47.9
Claude Sonnet 4.6
+8.9
Stability: 29.6 → 38.5
GPT-5.5
+6.5
Stability: 34.2 → 40.7
GLM-4.6
+5.0
Value: 32.7 → 37.7
Gemini 2.5 Pro
+4.4
Stability: 31.2 → 35.6
GPT-o3
+3.9
Stability: 36.1 → 40.0
Qwen3 Max
+2.5
Value: 44.3 → 46.8
Gemini 2.5 Pro
+2.4
Value: 33.9 → 36.3
Claude Sonnet 4.6
+1.9
Value: 24.4 → 26.3
Grok 4
+1.6
Value: 24.5 → 26.1
GPT-5.5
+0.7
Value: 18.6 → 19.3
Qwen3 Max
+0.5
Stability: 21.6 → 22.1
Doubao Pro
-26.8
Stability: 75.0 → 48.2
DeepSeek V4 Pro
-11.0
Availability: 99.0 → 88.0
DeepSeek V4 Pro
-1.3
Value: 44.3 → 43.0
Claude Opus 4.7
-1.0
Stability: 45.4 → 44.4
GPT-o3
-1.0
Availability: 98.0 → 97.0
GLM-4.6
-0.8
Availability: 67.1 → 66.3