AI on 28 Oct 2026
Everything new on the trackers that day: official announcements, model uploads from the big labs, tool releases, papers, AI rules and outages. Each item links to its source.
On 28 Oct 2026 the Machine Notes trackers recorded 1 announcement, 1 new model, 1 tool release, 5 papers and 1 outage (announcements from Meta). The most upvoted paper was "ChronoReplay: One Answer Interface Is Not Enough for Streaming Spatial Intelligence". The fastest-rising AI repo was AienTech/dossier.
Official announcements
WireNew models from the big labs
Models| Model | Lab | Task | Downloads | Uploaded |
|---|---|---|---|---|
| LLMON-Benchmarksibm-granite | IBM | – | 0 | Oct 28 |
Tool releases
Releases that day| Tool | Release | Published |
|---|---|---|
| llama.cppggml-org/llama.cpp | b11946Pre-release | 01:58 UTC |
Papers people are reading
Research that day| Paper | Upvotes |
|---|---|
| ChronoReplay: One Answer Interface Is Not Enough for Streaming Spatial IntelligenceWhen post-training emphasizes one readout, improvements can therefore concentrate on that interface while transfer to others remains weak or even degrades; we call this observed specialization interface lock-in.Yihao Hu, Yitong Lin, Pan Wang and others | 126 |
| Hy-Nomas: Towards Free-form Interactive World GenerationWe present Hy-Nomas, a video world model with an 8B diffusion backbone that supports free-form text instructions and continuous camera control across multiple interaction turns.Qiaoling Chen, Jiaxiang Cheng, Xueqing Deng and others | 71 |
| Rollback: Experience Backtracking in World Models for Policy ImprovementWe introduce Rollback, a policy improvement paradigm that turns world models from passive simulators into counterfactual engines: it rolls back to critical decision points, branches into alternative actions, and contrasts their outcomes to attribute success to individual…Jiayi Luo, Yang Zhang, Liliang Chen and others | 43 |
| ThinkWorld: World Imagination Reinforces Chain-of-Dynamics ReasoningWe introduce ThinkWorld, a VLA learning framework that grounds action reasoning in world dynamics and strengthens it through reinforcement learning with world imagination.Chaehyun Kim, Chi-Pin Huang, Jisoo Kim and others | 41 |
| Representation Redistribution in Sparse AutoencodersWe introduce the redistribution sensitivity score (RSS), which identifies large feature responses that recur in a consistent direction, and evaluate the resulting supports on variants excluded from selection.Ziye Mei, Haotian Ma | 36 |
Outages
Outage log| Service | Incident | Impact | Started | Lasted |
|---|---|---|---|---|
| Anthropic | Elevated errors affecting the Gmail connectorclaude.ai · Claude Code · Claude Cowork | Minor | 03:59 UTC | 1 h 49 min |
Rising AI repos
Repos that day| Repository | Description | Stars gained |
|---|---|---|
| dossierAienTech | A self-hosted job search app that writes tailored resumes and cover letters from your profile. | +1,175 |
| anymovepatch-gg | Motion capture for any creature from one AI video: two cameras in one frame, a DINOv2 joint reader, the rig fitted in both views, a GLB out. | +187 |
| openqodexopenqodex | Open source AI code review for coding agents. A separate reviewer checks every change your agent writes before you push, in Claude Code, Codex, Cursor and Cline. No account or key of its own. | +102 |
| earshotLeonxlnx | Ears for coding agents: let AI hear, review and make sound. CPU-only, open source. | +84 |
| project-mayamw00 | Project Maya - run GLM-5.3-Flash (321B MoE), DeepSeek-V4.1-Flash-1017 (748B MoE) and MiniMax-M3.1-Flash (428B MoE) privately on your own GPU(s), from one card to 16, NVIDIA or AMD, Linux or Windows. Engine, server, dashboard, OpenAI- and Anthropic-compatible API, quant tools. | +77 |