Crimson AI NewsA CrimsonLingua Network service
EN ع
← Back to news
Research paper Hugging Face

VibeWorlding: RL-Trained Open-Source Agents Outperform Closed-Source in 3D World Building

AI By Crimson AI Hugging Face Papers 18 August 2026 · 00:00 13 views
Share: X Telegram

Hugging Face researchers introduce VibeWorlding, a unified framework that benchmarks and trains multimodal agents to construct 3D open worlds end-to-end, showing that reinforcement learning enables open-source models to surpass closed-source frontiers.

VibeWorlding: RL-Trained Open-Source Agents Outperform Closed-Source in 3D World Building

Key points

Constructing an interactive 3D open world from a simple user query is a complex task that requires understanding intent, planning scenes, and using 3D tools. However, existing methods are often evaluated on idealized queries, making it hard to compare how multimodal agents handle real-world requests. To address this, Hugging Face researchers propose VibeWorlding, a unified framework for benchmarking and training 'vibe worlding agents'—multimodal agents that autonomously infer user intent, plan layouts, invoke 3D tools, and reflect on feedback in a multi-turn interaction.

The framework includes VWE-BENCH, a benchmark with 2,616 high-quality 3D assets, 323 human-annotated seed worlds, and 6,828 reverse-synthesized queries, split into verified and unverified sets. It also introduces VibeWorlding-Gym, a joint multimodal RL post-training framework that integrates a sandbox environment (with asset retrieval, editing, and rendering as MCP tools) and a rubric-based verifier for physical feasibility and intent fulfillment.

Experiments reveal that even frontier models like GPT-5.5 and Qwen3.8-Max achieve below 60% success rate, with the bottleneck being precise 3D world editing. However, RL training eases this weakness: the open-source VibeWorlder-8B matches frontier models, while VibeWorlder-30B-A3B attains the best overall Pass@1 among all evaluated models.

This work provides the first agentic RL framework for end-to-end 3D world construction, offering a unified stack of assets, benchmarks, and training methods. It highlights the potential of open-source models to surpass closed-source systems when trained with reinforcement learning.

ModelSuccess Rate
GPT-5.5<60%
Qwen3.8-Max<60%
VibeWorlder-8BComparable to frontier
VibeWorlder-30B-A3BBest overall Pass@1
Source
Hugging Face · Hugging Face Papers
Related news
Research paper
Hugging Face 29 Aug 2026

Hugging Face Audit: 110 of 124 AI Evaluations Fail to Support Their Claims

A new commit-bound census of 124 Inspect Evals units reveals that 110 stop before deterministic inference due to missing historica...

4
Research paper
Hugging Face 29 Aug 2026

Aphanta: New Framework Diagnoses When Image Editing Boosts Multimodal Reasoning

Hugging Face researchers introduce Aphanta, a diagnostic framework that evaluates when image-editing intermediates improve multimo...

5
Research paper
Hugging Face 29 Aug 2026

EditaLive! Enables Real-Time Character Video Editing for Live Streaming

Hugging Face researchers introduce EditaLive, a framework for real-time human-centric video editing in live streams, achieving sta...

4