SpatialWorld - Travel: leaderboard

Metric: Task success rate (%) on the 132 travel tasks (mostly CARLA driving and EmbodiedCity aerial navigation); task success rate (%) judged by terminal-state verifiers, vision-only egocentric observation and a unified text action interface, temperature 1.0, 30-turn history, step budget twice the human golden action count plus 10; higher is better. Source: arxiv.org. Saturation forecast: Around 2030. 15 models tracked.

Top models

#ModelScore
1GPT-56.8
2Gemini 3 Flash6.1
3Gemini 3.1 Pro (Preview)4.5
4Qwen 3.5 397B A17B4.5
5Qwen 3 VL 235B A22B Instruct4.5
6Gemini 2.5 Pro3.8
7GPT-5.43.8
8Kimi K2.53.8
9Qwen 3 VL 235B A22B (Thinking)3.8
10Seed 2.0 Lite3
11GLM-4.6V1.5
12GLM-4.5V1.5
13Qwen 2.5 VL 72B Instruct0.8

Interactive version: theaggregate.ai/benchmark?slug=spatialworld-travel · How It Works · Data refreshed daily, snapshot 2026-09-29.