Ant Group’s LLaDA2.2-flash edits its own generations and claims 1.64× speed
dr_cintas · x · 2026-07-25
- Ant Group reportedly released LLaDA2.2-flash, a diffusion model that can revise its own output during generation.
- The model is described as a real agent that can plan and call tools while keeping block-parallel speed.
- Claimed results:
- Better than the top autoregressive model on τ²-Bench and MCP-Atlas
- 1.64× higher throughput
- Native 128K context
- The post says it is 100% open source.
More from Models
- Benchmark scores drop from 89% to 19% on new evals — how benchmaxxing breaks leaderboard trust — airesearch12 · 2026-09-11
- Developer Building a Unified Leaderboard of All Model Benchmark Scores — airesearch12 · 2026-09-11
- Rumor claims Kimi faked performance by serving Claude; DeepSeek new model surprises in evals — realsohamparekh · 2026-09-11
- GPT-5.6 writes well but is instantly forgettable, user complains — BasedRaddka · 2026-09-11
- Opus Refuses Protein Research Codebase Over 'Safety' Concerns, Dev Considers Rolling His Own — josephdviviano · 2026-09-11
- User Hails Unconfirmed 'DeepSeek 4.1 Flash' as an Inflection Point in LLMs — himanshustwts · 2026-09-11