Cola2: ByteDance's latent-span language model that writes in continuous space
ByteDance Seed released Cola2, a 9B continuous-latent language model that groups text into semantic spans, compresses each into a compact latent, generates latent blocks with a diffusion transformer, and decodes them into tokens; the reported checkpoint trained on 346B tokens averages 60.10 across eight tasks.
Why it matters
It is a genuinely different generation architecture: reasoning and generation happen over compressed semantic spans rather than token by token, and the inference-only code is Apache-2.0 with the paper on arXiv.
What you could build with it
A research team studying cheaper long-form generation could use Cola2's latent-span approach to prototype a summarizer that reasons over compressed semantic spans instead of raw tokens, cutting the compute spent on very long documents.
Does it hold up?
Too early to judge: the repo went public today with inference-only code and the 9B checkpoint has not been released yet, so nobody outside ByteDance has run it.
Learn more
First spotted on github: source.
More AI models
VoiceStudio: open-source, fully-local ElevenLabs alternative for voice cloning, dubbing, and audiobooksdebpalash/VoiceStudio is a desktop app that runs voice cloning, voice design, video dubbing, dictation, transcription…models · JEV 0.75Soup: fine-tune an 8B LLM on a 4 GB laptop GPU from one YAML fileAn open-source CLI that fine-tunes LLMs from a single YAML file, with layer streaming that trains an 8B model on a 4…models · JEV 0.73Tencent Hunyuan Hy3: open 295B MoE with 21B active under Apache 2.0Tencent open-sourced Hy3, a 295B-parameter mixture-of-experts model with only 21B parameters active per token, under…models · JEV 0.73NVIDIA Isaac Lab 3.0 GA + Isaac OS open runtime + Agile ONE SN humanoidMajor Isaac platform expansion at GTC Berlin Oct 21: Isaac Lab 3.0 is generally available as the open-source robot…models · JEV 0.73NVIDIA Nemotron 3.5 Retriever collection: Embed 8B/1B, Rerank 8B, Parse 631MNVIDIA's Oct 21 collection drop for agentic retrieval: Nemotron-3.5-Embed in 8B BF16, 1B BF16 and 1B NVFP4…models · JEV 0.72Kimi K3.1 Pro: Moonshot's 2.8T open-weight flagship officially launchesMoonshot AI officially launched Kimi K3.1 Pro on Oct 23, a mixture-of-experts model with 2.8 trillion total / 104…models · JEV 0.72
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.