STEALTHMARK INTELLIGENCE BENCHMARKS
Where Qwen 4 could land
Read the Qwen 4 guide ↗
Compare measured capabilities across our benchmark tasks. Explore models and families, or add your own estimate.
Qwen4-Max (projected): StealthMark Score 58.8, General Intelligence 54.6, Visual Intelligence 66.3.
| Model | StealthMark Score | General Intelligence | Visual Intelligence |
|---|---|---|---|
| Qwen4-Max · Editorial estimate | 58.8 (Central 58.8–60.8; Wider 58.8–62.8) | 54.6 (Central 54.6–56.6; Wider 54.6–58.6) | 66.3 (Central 66.3–68.3; Wider 66.3–70.3) |
| Qwen4-Plus · Editorial estimate | 57.5 (Central 56.0–60.0; Wider 54.5–62.5) | 51.5 (Central 50.0–54.0; Wider 48.5–56.5) | 67.9 (Central 66.4–70.4; Wider 64.9–72.9) |
| Qwen4-Flash · Editorial estimate | 56.2 (Central 53.2–59.2; Wider 50.2–62.2) | 48.3 (Central 45.3–51.3; Wider 42.3–54.3) | 69.5 (Central 66.5–72.5; Wider 63.5–75.5) |
| Qwen4-27B · Editorial estimate | 61.5 (Central 58.5–64.5; Wider 55.5–67.5) | 56.1 (Central 53.1–59.1; Wider 50.1–62.1) | 71.0 (Central 68.0–74.0; Wider 65.0–77.0) |
| Qwen3.8-Max-0902 | 62.4 | 61.6 | 63.9 |
| Qwen4-Max-Preview | 58.8 | 54.6 | 66.3 |
| Qwen3.8-27B | 51.5 | 46.1 | 61.0 |
| Qwen3.8-Flash | 46.2 | 38.3 | 59.5 |
| GPT-6 Astra | 74.6 | 73.4 | 76.9 |
| Claude Opus 5.5 | 72.6 | 71.3 | 75.0 |
| Claude Sonnet 5.5 | 70.2 | 67.0 | 76.1 |
| GPT-6.1 Sol | 67.8 | 64.4 | 74.1 |
| Claude Haiku 5.5 | 49.5 | 47.6 | 52.8 |