Claude agents discover novel enzyme system as Alibaba starts Qwen4 training

The architecture targets agentic workloads, where each turn appends long observations and prefill cost, cache size and retrieval accuracy sit on the critical path simultaneously. Switching from block-level to token-level selection and folding SWA into a forced recent window gives local and global tokens one shared KV cache. Developers building long-context agents should watch MiMo-V3's release, especially against open-source alternatives like DeepSeek and Qwen in the same weight class.

Putting real-time generation and general world-modelling tasks in one model could cut the latency and engineering cost of running separate specialised systems. Teams building interactive video, simulation or embodied applications should track how R2 performs against pipelines that bolt a world model onto a streaming engine. PixVerse competes with video generation tools like Runway and Kling in the same product space.

量子位@梦瑶Source

The stack signals Alibaba Cloud is positioning infrastructure, not just models, as the battleground for agent workloads — governance, storage throughput and networking are tuned for large-scale agent fleets. Cloud architects and platform teams building agent deployments in Southeast Asia can benchmark these specs against AWS Bedrock AgentCore and Google's agent offerings. Chinese cloud providers' regional data-centre presence makes this stack directly available to ASEAN enterprises.

官方一手pandaily@[email protected] (Pandaily)Source

The release joins a crowded field of open-source coding agents from Chinese labs, including MiniMax Code, Zhipu's ZCode, Qwen Code, Kimi Code and DeepSeek's DSH, giving developers more MIT-licensed CLI options to self-host or benchmark. Since the Multi-Frame score comes from StepFun's own benchmark, cross-checking it against Terminal-Bench results from rival agents is the practical way to compare. The 75-day Plus subscription also offers a low-cost way to test Step 5 Preview before committing.

shao__meng@shao__mengSource

The method shows how graph neural networks can extract clinically meaningful subtypes from population-scale single-cell data, turning donor tissue into a map of resilience and psychiatric comorbidity in AD. Researchers in computational biology and drug discovery can reuse the approach on other diseases where donor cohorts exist. Teams building biomarker pipelines should note the link between cell-level patterns and depression onset in AD patients.

Nature Medicine@Daifeng WangSource

The practical takeaway is that effort settings do not transfer across models: teams reusing old configurations for Opus 5.5 may overspend or trigger refusals from leftover reasoning-extraction instructions. The checklist-based harness pattern for preventing premature end_turn stops is directly reusable in long-running agent workflows. Developers running unattended coding, audit or migration tasks on the Claude API will get the most value from the guide.

shao__meng@shao__mengSource