AI Coding Live News

Modern video foundation models synthesize photorealistic visual dynamics, yet struggle on long-horizon procedural tasks requiring sequential causal execution—such as equipment maintenance, cooking, or laboratory experiments. Current architectures rely on open-loop generation that cannot adapt to intermediate video outputs, resulting in disordered task sequences, hallucinated progress, and premature task termination. Researchers from MBZUAI and the Australian National University introduce WorldGuide (arXiv:2610.12459, repository: github.com/mbzuai-oryx/WorldGuide, site: mbzuai-oryx.github.io/WorldGuide/), formalizing procedural video generation as closed-loop task execution in visual world space. Given solely an initial image and a high-level task goal, WorldGuide iteratively predicts the next atomic action, generates the corresponding visual video clip, and inspects the synthesized state to dynamically route subsequent actions or determine task completion. A hierarchical visual memory maintains state history under bounded token footprints. To bridge supervision gaps, the authors release WorldGuide-Bench comprising approximately 59K step-annotated videos across 245 tasks and 27 procedural categories. WorldGuide achieves 33.33% Task Success on WorldGuide-Bench, outscoring the strong MiniMax-H3 model (29.90%) even when MiniMax-H3 is provided with reference plans, and surges to 47.69% on VideoCraft-Bench compared to 32.73% for MiniMax-H3 under goal-only conditioning.

On Oct 7 Perplexity released pplx-embed-v2-late, multimodal late-interaction (ColBERT) retrievers in 0.6B (340M active) and 9B (7.4B active) sizes, built on Qwen3.5 with bidirectional attention. Each token gets a 128-dim vector scored by MaxSim, covering text, images and rendered PDFs or slides without OCR. Both sizes share one embedding space, so the 0.6B model can query a 9B-built index. Public ViDoRe v3 image nDCG@10 is 62.3% / 65.2%; MIT-licensed weights are on Hugging Face.

Community coverage notes Alibaba open-sourced its internal AI code-review assistant as Open Code Review (github.com/alibaba/open-code-review; npm @alibaba-group/open-code-review at 1.12.0). The CLI reads Git diffs, uses a tool-using agent for line-level structured comments, supports ocr scan for whole-file audits, and plugs into Claude Code, Codex, and Cursor. Official bench: 50 repos / 200 PRs / 10 languages with 1,505 annotated issues; same base model yields higher Precision and F1 than general-purpose Claude Code at ~1/9 the tokens, trading some recall for less noise.

Open-source dsh-ios (DeepSeek Harness plugin) wires the iOS Simulator and USB devices into the chat: 22 tools cover build, boot, logs, OCR, accessibility tree, and UI control, with a live sidebar for tap/drag. Device signing follows configurable Xcode teams and stays local with short-lived signatures; requires macOS and full Xcode (github.com/ZSeven-W/dsh-ios).

LatentPress stores conversational histories and long documents as continuous memory tokens a frozen decoder reads through its input-embedding interface—no text reconstruction at inference. A small reader-matched writer (4.2M–26.2M adapter params, ~0.1% of the decoder) compresses 4–16×. On LongMemEval it reaches 0.504 accuracy at 7.70× compression vs 0.490 uncompressed, beating text summaries and OCR; writing is ~43ms per conversation and reading is 5–9× faster than raw context—useful for long-context agent memory.

Cohere launched Parse 5, a document intelligence model for high-volume enterprise RAG and document agents scoring 79.2 on ParseBench (beating Mistral OCR 4 and Azure) priced at $1.50 per 1,000 pages.