New spatial planning benchmark: GPT-6 and Claude Opus 5.5 fail parking route tasks
songhan_mit · x · 2026-11-22
Researcher Jinglun Zhao built a spatial planning evaluation for LLM agents using parking as a testbed, covering 6 layouts (offset-entry, dead-end, T/L-shaped aisles, angled bays, curb extensions) scored on distance, gear changes, and clearance. Key findings: manipulation tasks are nearing saturation, but spatial planning still separates models sharply — GPT-6-Astra failed to park on the first two layouts every time, Claude Opus 5.5 never found a valid route on either, and among valid routes quality varied wildly (scores 78.8 vs 11.0, worst clearance 5.90 cm). Routes were validated with full-body collision checks from recorded trajectories.
More from Models
- Claude veteran shocked by Gemini's leap, weighs switching to Google AI Ultra — ScreenPlayLife · 2026-11-22
- Gemini Neon builds an interactive USL lock-contention UI exposing a testing blind spot — tianyin_xu · 2026-11-22
- OpenAI details how it detects and fights adversarial model distillation — rhiever · 2026-11-22
- Microsoft's new models reportedly up to 80x cheaper and 50x faster — ChrisUniverse · 2026-11-22
- Inkling 2 Mini uncensored: refusals 100%→0% with a 40MB LoRA, 98% MMLU kept — Top-Eye-8104 · 2026-11-22
- GPT-6.2 Astra writing success rate falls to 85.9% from 91.2% — arena · 2026-11-22