PerspectiveGap - Combined: leaderboard
Measures whether a model can compose role-specific prompts for a multi-agent orchestration without over-sharing context. Each of 110 scenarios supplies a role list, a shuffled set of information fragments and one injected distractor, against a reference assignment giving each role exactly the fragments it needs. A deterministic rule-only scorer, not an LLM judge, audits containment using phrase-level fragment fingerprints. Strict pass is all-or-nothing: an evaluation scores 1 only when it omits no required fragment and includes nothing out of role. Combined is the unweighted average of the strict pass rates of the two task formats. This series reports PerspectiveGap - Combined using Pass Rate (%).
Metric: Pass Rate (%). Source: huggingface.co. Saturation forecast: Around December 2027. 35 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 62 | #33 |
| 2 | GPT-5.6 Terra | 42.7 | #40 |
| 3 | GPT-5.6 Sol | 42.5 | #22 |
| 4 | Grok 4.6 | 40.9 | #29 |
| 5 | DeepSeek V4 Pro | 32 | #112 |
| 6 | Claude Fable 5 | 31.4 | #20 |
| 7 | Kimi K2.6 | 25.9 | #113 |
| 8 | Claude Sonnet 5 | 25.7 | #65 |
| 9 | DeepSeek V4 Flash | 24.8 | #155 |
| 10 | Gemini 3.1 Pro (Preview) | 21.8 | #64 |
| 11 | GLM-5 | 20.2 | #149 |
| 12 | MiMo-V2.5-Pro | 20 | #167 |
| 13 | Claude Opus 4.7 | 19.1 | #51 |
| 14 | Qwen 3.6 Plus | 17 | #124 |
| 15 | GPT-5.6 Luna | 16.6 | #79 |
No result here: #3 Claude Opus 5.5, #7 GPT-6.1 Sol, #9 Gemini 4 Argon.
Interactive version: theaggregate.ai/benchmark?slug=perspectivegap-combined · How It Works · Data refreshed daily, snapshot 2026-11-05.