VitaBench 2.0: leaderboard
Large language models (LLMs) have evolved into interactive agents that collaborate with users in real-world tasks. Effective collaboration in such settings increasingly depends on understanding the user beyond what is explicitly stated, as user intent is often reflected in fragmented daily interactions and requires both personalized modeling and proactive interaction. However, existing agent bench
Metric: Avg@4 Full Context (imported). Source: benchmarklist.com. Saturation forecast: Around March 2027. 45 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.6 | 50.3 | #63 |
| 2 | Claude Opus 4.6 (Thinking) | 50.3 | #63 (Claude Opus 4.6) |
| 3 | Kimi K2.6 (Thinking) | 48.1 | #104 (Kimi K2.6) |
| 4 | Seed 2.0 Pro | 47.4 | #118 |
| 5 | Seed 2.0 Pro (Thinking) | 47.4 | #118 (Seed 2.0 Pro) |
| 6 | DeepSeek V4 Pro | 47.2 | #102 |
| 7 | DeepSeek V4 Pro (Non-reasoning) | 45.6 | #102 (DeepSeek V4 Pro) |
| 8 | GLM-5.1 (Thinking) | 45 | #97 (GLM-5.1) |
| 9 | GPT-5 | 44.1 | #98 |
| 10 | GPT-5 (Thinking) | 44.1 | #98 (GPT-5) |
| 11 | Seed 2.0 Pro (Non-reasoning) | 42.8 | #118 (Seed 2.0 Pro) |
| 12 | GLM-5.1 | 42 | #97 |
| 13 | GLM-5.1 (Non-reasoning) | 42 | #97 (GLM-5.1) |
| 14 | Claude Sonnet 4.5 | 41.7 | #142 |
| 15 | Claude Sonnet 4.5 (Thinking) | 41.7 | #142 (Claude Sonnet 4.5) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=vitabench-2-0 · How It Works · Data refreshed daily, snapshot 2026-10-20.