University of Illinois Urbana-Champaign

Trends

• Agent benchmarks expose reliability gaps and lose trust
• Open-weight models diversify in architecture, modality, and size

Items

1. Oct 2

   UIUC's hierarchical continuous diffusion LM

   HC-DLM makes a continuous latent trajectory the only persistent generative state, reading tokens out at each step and feeding them back as scaffolding, and beats discrete and continuous diffusion baselines on Sudoku, Countdown, and LM1B at matched size.

2. Jun 21

   UIUC's PlanBench-XL

   An interactive benchmark of 327 retail tasks over 1,665 tools with optional missing, failing, or distracting tools; GPT-5.4 falls from 51.90% without blocking to 11.36% under severe blocking, exposing brittle long-horizon planning when failures lack explicit errors.

3. Jun 8

   Harness-1 open search agent beats GPT-5.4

   A 20B open-source search agent from UIUC, UC Berkeley, and Chroma, built on gpt-oss-20B, redesigns complex retrieval to reach 73% average recall versus GPT-5.4's 70.9%, with weights and environment released under Apache 2.0.

4. May 28

   UIUC's CausaLab tests causal discovery

   Agents must recover the causal model behind a synthetic lab; GPT-5.2-high reaches 92% task accuracy but only 0.471 all-edge F1 on the causal graph, with premature stopping as the main failure mode.