LLaDA2.2-flash brings 100B MoE diffusion model with 128K context and Apache 2.0
QuixiAI · x · 2026-07-24
inclusionAI released LLaDA2.2-flash, a 100B MoE diffusion language model aimed at agentic workloads.
The model supports a 128K context window, uses Levenshtein Editing for insert/delete-style parallel decoding, and adds Block Routing plus L-EBPO to improve long-context efficiency. According to the post, it beats Ling-2.6-flash on τ²-Bench, PinchBench, and MCP-Atlas, while delivering 1.7×–2.3× higher throughput across reported tests. The release is under Apache 2.0.
Related event: inclusionAI Launches 100B MoE Diffusion Language Model LLaDA2.2-flash(2 posts)→
More from coding & agent
- User says Claude gave up in 2 minutes, Codex felt much better for coding — jxnlco · 2026-07-24
- Users say GPT-5.6 Pro is better at reviewing Codex code than Codex itself — jarrodwatts · 2026-07-24
- Conductor team usage data shows create-pr and debugging dominate its workflow — charlieholtz · 2026-07-24
- Codex voice mode could let developers steer coding tasks hands-free — jxnlco · 2026-07-24
- Users are turning Codex voice mode into a phone-based coding feedback loop — jxnlco · 2026-07-24
- Codex voice mode is getting called “JARVIS at home” by users — jxnlco · 2026-07-24