Metraipolis › archive › dc_b5fc098db75e
DrivingBench: Can Vision-Language Models Drive a Toyota Corolla? — arXiv:2609.38948
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.38948
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T12:44:37.312Z
- SHA-256 of the archived text
e7a4568ac66a4bbb9a208039ddd940b955094aafc0584364ebd6d08f66a6977d- Stable URL of this page
https://metraipolis.com/documento/dc_b5fc098db75e
How to cite
DrivingBench: Can Vision-Language Models Drive a Toyota Corolla? — arXiv:2609.38948 — arXiv (en), https://arxiv.org/abs/2609.38948 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_b5fc098db75e (Metraipolis archive dc_b5fc098db75e, sha256 e7a4568ac66a4bbb9a208039ddd940b955094aafc0584364ebd6d08f66a6977d)
BibTeX
@misc{dcb5fc098db75e,
title = {DrivingBench: Can Vision-Language Models Drive a Toyota Corolla? — arXiv:2609.38948},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.38948},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_b5fc098db75e; sha256 e7a4568ac66a4bbb9a208039ddd940b955094aafc0584364ebd6d08f66a6977d},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.38948. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Frontier models excel at many digital benchmarks, yet their ability to drive a real car, an everyday human skill, remains largely untested. We present DrivingBench, to our knowledge the first benchmark where general-purpose vision-language models must drive a real car. Through three tools, the models see camera frames from a Toyota Corolla and directly command its steering and velocity around a parking lot cone course at low speeds. The car may continue moving while the model thinks and new commands replace the currently running one, so inference latency is part of the task, testing the models' abilities to observe, act, monitor, recover, and complete a long-horizon objective under such constraints. We benchmark GPT-6 Astra, Claude Fable 5.1, GPT-5.6 Sol, and Grok 4.6 in vendor-native harnesses (Codex, Claude Code, Cursor) with up to three attempts each in one conversation; Astra is the only model to finish the course, on its second attempt, with no other attempt passing 50% of the course. Two of the four models improved materially across attempts with retained context. We also detail the design principles behind our action interface, and show how the tool output format and the framing of the task combined to determine whether models would drive at all or refuse. We release our harness, prompts, course map, and traces with video and telemetry for reproducibility. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-01 in cs.AI). Categorie arXiv: Robotics (cs.RO) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2609.38948 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-07T12:14:13.982Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).