Fable 5.1 배포 확대, Atlas의 3D 카메라 제어, 공개 영상 생성과 에이전트 메모리 설계

Claude Fable 5.1, the latest in @AnthropicAI 's Mythos model class, is now generally available in GitHub Copilot. In our testing it showed strong performance on long-running coding tasks deep codebase research complex agentic workflows Try it out in the GitHub Copilot app, CLI or @code . https:// github.blog/changelog/2026 -09-01-claude-fable-5-1-generally-available-in-github-copilot/ …

Claude Fable 5.1 from @AnthropicAI is now available on DigitalOcean Inference Engine, day zero. https:// do.co/4yiRcZK Cache-reads now cost 75% less, cutting the cost of long-running agents that lean on repeated context. Access it via Serverless Inference, Inference Router, model synthesis, and Evaluations.

Another glimpse of Real2Sim with Atlas: capture a real environment with just a number of casual photos from a phone, turn them into a sim, then simulate diverse robots navigating different trajectories. Atlas generates the RGB and depth observations they would see along the way.

Omg, what just dropped? Atlas AI can now build a 3D world and let you keyframe the camera. Filmmaking just changed. Pixel-perfect camera control in generated 3D worlds. This is the feature filmmakers have been waiting for. From prompt to camera move. Atlas is the first model that actually lets you direct space, not just describe it. https:// x.com/EHuanglu/statu s/2094867317658472506/video/1 …

This is what open-source intelligence is for. Built on vLLM-Omni and @haoailab 's FastH3, with @NVIDIAAI hardware support - all public. Real-time generation makes interactive video possible; an open baseline makes it improvable for everyone. Take it and go faster!

We would never see this level of creation everywhere if SOTA video models stayed behind closed doors. Every single day, we're blown away by what you're building with MiniMax H3. Remind yourself: MiniMax H3 open weights dropped just one month ago. Imagine where we'll be in a year.

Super interesting paper from Meta. Long-horizon research agents are coming. But one common problem with research agents today is the lack of originality and how to decide what experiments are worth exploring. An AI research agent can propose far more experiments than it can afford to run, so the problem is not idea generation, it's deciding which candidates get GPU time. AI Research Preference Models is trained to predict which candidate solution is most promising before any of them execute. Two variants, both built on frozen pretrained LLMs. An inference-only model reasons over candidate plans, code, and previously executed solutions. An agentic model additionally runs small-scale pilot experiments before committing budget. Dropped into the AIRA-dojo agent and measured on AIRS-Bench, average normalized score moves from 0.684 to 0.711 and 0.729. Both variants reach the unguided agent's 24-hour performance in roughly 15 hours, using less than two-thirds of its execution budget, and together set new state of the art on two AIRS-Bench tasks. Paper: https:// arxiv.org/abs/2608.13940 Chat with Paper: https:// academy.dair.ai/papers/ai-rese arch-preference-models-2608.13940 …

// Agent Zero Memory // This work separates three things that agent memory systems usually collapse into one. If you build agents with long-term memory, this memory design is a worth a read. Here is how it works: Agent Zero Memory runs an episodic events timeline, an entity-event knowledge graph, and a curated documentary memory of durable facts side by side over the same history. A retrieval turn passes through an intent gate, then a source router, then three concurrent agentic searches, one per system. Every stored item carries its origin, timestamp and evidence pointer, and answers run under a citation lock, so a reply may cite only evidence its reader actually opened. When the evidence is missing the system abstains. It reaches 95.60% on LongMemEval and 93.60% on LoCoMo, both new highs. The cost result is the more useful one for AI builders. Across eight backbone models accuracy moves by 3.4 points while per-query cost moves about 30x, with near state of the art quality available at up to 20x lower cost per query. Memory design is driving the quality here. Paper: https:// arxiv.org/abs/2608.29606 Chat with Paper: https:// academy.dair.ai/papers/agent-z ero-memory-provenance-aware-long-term-memory-for-llm-agents-2608.29606 …

Sydney Runkle

@sydneyrunkle

4시간 전

you can now build @LangChain agents w/ agentic video understanding! google is consistently at the forefront of multimodal processing; this is the video analog of "agentic vision" token usage + cost is also way down w/ this new approach! docs: https:// docs.langchain.com/oss/python/int egrations/chat/google_generative_ai#agentic-video-understanding …