Inclusion AI ships LLaDA2.2-flash, a 1.6x faster diffusion model for agents
heyshrutimishra · x · 2026-07-25
Inclusion AI has released LLaDA2.2-flash, a large-scale diffusion language model aimed at agentic workloads.
The post argues that most agents still rely on autoregressive models that generate one token at a time, which was built for conversation rather than planning, tool use, and self-correction over long trajectories. LLaDA2.2-flash instead generates tokens in parallel blocks and uses Levenshtein Editing to delete defects and insert fixes mid-sequence.
According to the post, the model is competitive with top autoregressive models on agentic benchmarks, runs 1.6× faster, supports a native 128K context window, and is open source. The author says this is the first diffusion model that looks viable for production agent workloads, and links a GitHub repo, Hugging Face page, and tech report.
More from coding & agent
- Reticle builds a verification layer that lets AI coding agents check their own generated code — Scobleizer · 2026-10-27
- Dev training a guard model to block prompt injection, tool and memory poisoning — gnukeith · 2026-10-27
- Dev trains a ~4GB-VRAM sentinel model to guard against prompt injection and poisoned memory — gnukeith · 2026-10-27
- Ex-Big Tech managers may thrive in the AI era: manage agents like you managed teams — lxfater · 2026-10-27
- Claude debugs airline and hotel sites; Fasano's date picker is all NaN and invalid dates — lilyraynyc · 2026-10-27
- SF 'Code Engineering' event picks up where 1,000-person harness engineering meetup left off — LangChain · 2026-10-27