• Readable Dynamics Don't Belong in Weights. Enterprise World Models use CascadeBench to show that cross-tenant business rules get more brittle the better they're learned. 58 upvotes are redrawing the line between RAG, tool calls, and parametric knowledge.

• AlphaGRPO Skips the Cold-Start Step for Unified Multimodal Models. Decomposing rewards into atomic verifiable questions (DVReward) lets GRPO unlock self-reflective refinement directly. GEdit gains transfer without ever training on the editing task.

• ToolCUA Trains on Trajectory Orchestration, Not Single Steps. OSWorld-MCP jumps from a 28% baseline to 46.85%, beating the pure-GUI setting by 3.9%. CUA fails on paths, not on clicks.

• L2P Drops VAE for Large Patch Tokens. Freeze a pretrained LDM as a prior extractor, train on synthetic data with 8 GPUs, get native 4K. GenEval cost: 93%.

• Async RL Silently Miscomputes the Importance Ratio. Training-inference discrepancy and policy staleness get tangled into one ratio, producing silent semantic mismatch. PPO-EWMA is the cheap fix.

Also Notable

• CHAS Hits the Other Side of CUA: Long-Tail Interaction Scarcity. Same day as ToolCUA, offers synthesis methods and a benchmark for complex, low-frequency GUI interactions. Covering Human Action Space
• Image Editing Benchmark Ships Alongside Reward Model Benchmark. Targets the ceiling for current frontier evaluation. Edit-Compass + EditReward-Compass form a unified framework. Edit-Compass
• Split Thoughts, Inputs, and Outputs Into Parallel Streams. Challenges the assumption that an agent must run on a single message sequence. Multi-Stream LLMs
• Tool-Using Agent Failures Happen at Trajectory Level, Not Final Response. Trajectory-level on-policy self-evolution sidesteps the classic safety-utility tradeoff. On-Policy Self-Evolution
• Convert a Pretrained LLM Into a Looped Latent Refinement Model. Test-time compute scaling doesn't require training a recurrent model from scratch. Existing LLMs can be reused. LoopUS
• World Prediction and Action Generation Couple Together. DAWN breaks the "predict-then-act" serial assumption. Maneuver and scene evolution condition on each other. DAWN
• Long-Horizon Agents Switch to "Map-Then-Act." Build an environment map first, then execute, rather than reactively inferring constraints on the fly. MAP
• Black-Box DoS Attacks That Induce LRM Overthinking. A hierarchical genetic algorithm triggers excessive thinking. Compute availability becomes a new attack surface for reasoning models. Inducing Overthink
• Speculative Inference Framework for Diffusion-Based VLA. Most steps skip full inference, making dVLA real-time deployment feasible. Realtime-VLA FLASH
• Planner and Simulator Co-Evolve to Solve Manipulation Data Scarcity. RoboEvolve sidesteps the semantic-spatial misalignment of VLM/VGM. RoboEvolve

Read the full edition →