LLaDA2.2 pushes diffusion LLMs into agentic work with 128K context and self-editing
alifcoder · x · 2026-07-25
LLaDA2.2 is pitched as a large-scale agentic diffusion model that goes beyond text generation into planning, tool use, and self-correction on long multi-turn tasks.
- It uses Levenshtein Editing with KEEP / SUBSTITUTE / DELETE / INSERT so the model can rewrite its own sequence instead of freezing early mistakes into context.
- L-EBPO trains the model to decide when and where to correct errors using environmental feedback, aiming to reduce long-horizon drift and cascading failures.
- The system adds native 128K context and Block Routing for more controllable long-context inference.
- On 7 agentic benchmarks, LLaDA2.2-flash averages 53.83 vs. 55.74 for Ling-2.6-flash, while delivering 1.64× average BF16 throughput across 11 workloads.
More from Models
- Opus 5 is claimed to jump ahead on three spreadsheet-agent benchmarks — surmenok · 2026-07-25
- Claude Opus 5 launches with strong coding feedback and mixed benchmark signals — Latent Space · 2026-07-25
- xAI cuts SuperGrok Heavy to $99 a month with a 67% three-month discount — ns123abc · 2026-07-25
- Why Bonsai 27B Is Missing From Artificial Analysis Rankings — ML-Future · 2026-07-25
- Qwen 3.8 Max Preview wins praise for planning, simplification, and product videos — RevealIndividual7567 · 2026-07-25
- Gemini users report error 1706 on old chats while new conversations still work — Keithstrombiller · 2026-07-25