On October 28, Zhiyuan released an upgraded version of its embodied intelligence simulation platform Genie Sim, focusing on improving soft-body simulation and adding model evaluation and reinforcement learning post-training features. In comparative tests covering 4 robot platforms and 22 tasks, the correlation between simulation and real-robot task scores reached 96.7%. Related code is planned to be updated on GitHub, with simulation assets and datasets to be opened to developers.
Qiming Venture Partners' portfolio company Wuxian Qionxin announced the release of RLinf v0.4, an open-source reinforcement learning framework for embodied intelligence, jointly developed with Tsinghua University and Beijing Zhongguancun Laboratory.
Artificial Analysis announced the launch of Intelligence Index v5, adding two new benchmarks (FrontierCode 2.0 and Terminal-Bench-Science) and upgrading three existing benchmarks, with 60% of the index based on private datasets for the first time. It also released Progress Index v1 to track model capability changes over time.
Cognition released FrontierCode 2.0, a private coding benchmark with 132 real open-source repository tasks scored by whether maintainers would merge the changes. Claude Opus 5.5 leads at 46%, while DeepSeek V4.1 Flash is the top open-weights model at 28%.
SAP announced a range of AI innovations for developers at its annual SAP TechEd conference, including upgrades to Joule Studio, Joule Work, and the Business AI platform, the new table model SAP-RPT-2.0 with anomaly detection and multi-table processing, a collaboration with HPE to deploy SAP-RPT on the International Space Station, and free access to Joule Studio for some customers until December 2026.
Microsoft released MAI-Image-2.7-Preview on October 27, a controllable image editing model that takes an original image and text instructions. It ranks second on Artificial Analysis's image editing leaderboard with 1,193 Elo, behind OpenAI's GPT Image 2.5 Sunburst (max). The preview currently offers image editing only; no API, pricing, or commercial release date has been announced.
On October 27, Liquid AI released the LFM3 family, including six models spanning language, vision, speech, and structured decision-making. Built on a 1.7-billion-parameter foundation with specialized versions around 2 billion parameters, the models can be customized on a single GPU and deployed on devices, private servers, or cloud. The text model API is priced at $0.08 per million input tokens and $0.20 per million output tokens.
On October 27, OpenAI announced that it is opening iCloud Mail and iCloud Calendar plugins to all ChatGPT plan users, covering Free, Go, Plus, and Pro tiers on web, desktop, and iOS. After user authorization, ChatGPT can read iCloud inboxes, manage emails, and create or update events in Apple Calendar based on email information.
Salesforce on October 27 detailed the next phase of its Agent Fabric platform, comprising Agent Evals (beta in November, general availability in January 2027), the API Evals tool, a custom agent scanner beta, and an evaluation-triggered agent kill switch planned for January 2027.
On October 27, 2026, Google DeepMind released Genie 4, its next-generation world model, in research preview. It generates more realistic interactive worlds, supports real-time audio generation and scene editing, maintains scene consistency for up to two minutes, and integrates more closely with Gemini models.