Свежие AI-новости, релизы моделей и исследовательские заметки. Русская навигация локализована, сами материалы могут оставаться на языке оригинала.
RSS-лентаПеревод новостей в работе
Русские переводы готовятся. Пока статьи ниже отображаются на английском — навигация и метаданные локализованы.
TheRouter v2.0 introduces a unified async paradigm across image, audio, and video generation. Long-running upstream calls move off the request thread, into a job-based flow with S3-backed artifacts and a single /v1/jobs/:id polling surface.
DeepSeek's API now accepts Anthropic SDK format at api.deepseek.com/anthropic — meaning Claude Code, the Anthropic Python/TS SDK, and any Anthropic-native client can now route requests to DeepSeek V4 models without an OpenAI wrapper.
Anthropic has acquired Stainless, the company that generates every official Claude SDK and MCP server tooling. For teams building multi-provider API pipelines, this reshapes SDK dependency risk, MCP server governance, and the pace of Claude API surface changes.
Moonshot AI releases Kimi K2.6 with state-of-the-art long-horizon coding, multimodal input (text, images, video), 256K context, and a fully OpenAI-compatible API — directly affecting how engineering teams route coding-agent workloads.
Alibaba launched Qwen-Image and Qwen-Image-Edit on DashScope as qwen-image-2.0-pro. For routing teams, these introduce a non-DALL-E async job pattern and a new model namespace that standard OpenAI-compatible proxies may not handle correctly.
Alibaba Cloud's new Qwen-MT turbo model arrives via OpenAI-compatible endpoints, but its translation controls live inside extra_body — a pattern that breaks any middleware that strips non-standard fields. Here's what routing teams need to watch.
Alibaba Cloud's Qwen3.7-Max claims the top spot on Terminal Bench 2.0, outperforms Claude Opus 4.6 on GPQA Diamond, and completes a 35-hour autonomous agent task on novel hardware. Here's the operator angle for teams routing Qwen traffic.
DeepSeek published an official integration guide for Claude Code, OpenCode, and OpenClaw — revealing a per-tier model routing pattern that operators can apply across any Anthropic-compatible gateway.
DeepSeek deprecates deepseek-chat and deepseek-reasoner on July 24, 2026. Routing configs must migrate to deepseek-v4-flash or deepseek-v4-pro. A new Anthropic-compatible endpoint also changes how Claude Code users wire DeepSeek into their agent workflows.
Alibaba releases Qwen3Guard, an open-weight safety guardrail model with real-time streaming token-level detection. For teams routing across multiple AI providers, it offers a provider-agnostic safety layer you can deploy at the gateway level.
OpenAI, Anthropic, and GitHub changed pricing models in the same week. List price gaps reach 92% depending on tokenizer behavior and usage patterns. Routing architectures are now essential for cost control.
OpenAI is now a C2PA conforming generator with Google SynthID watermarks in every API-generated image. Every output carries a cryptographic provenance fingerprint — affecting audit trails, fallback routing, and enterprise content governance.
GSPO introduces sequence-level optimization for RL training, stabilizing MoE models and eliminating infrastructure-heavy Routing Replay workarounds.
A reported US review process for frontier AI models would make provider availability, release timing, and governance evidence part of routing strategy—not just policy news.
AI-generated images of colorful Audemars Piguet Royal Oak wristwatches flooded social media for a week before Swatch's actual Royal Pop pocket watch collaboration was revealed, creating unprecedented fake product hype.
OpenAI sought to present a jackass trophy in the Musk v. Altman trial as evidence of Elon Musk's behavior toward employees who challenged his safety priorities in 2018.
OpenAI CEO Sam Altman testified in the Musk v. Altman trial, revealing Elon Musk's suggestion to pass OpenAI control to his children and defending against allegations of deceptive behavior.
WhatsApp introduces Incognito Chat for Meta AI conversations, using Private Processing to prevent Meta from accessing user interactions with the chatbot.
The Musk v. Altman trial closing arguments reveal how OpenAI's nonprofit structure conflicted with competitive pressures, raising questions about public interest protection.
Sasha Luccioni launches Sustainable AI Group to address growing enterprise and regulatory demands for AI emissions data and energy efficiency metrics across European and global markets.
OpenAI confirmed two employee devices were hit by the Mini Shai-Hulud npm supply chain attack, exposing code-signing certificates for Codex CLI, ChatGPT Desktop, and Atlas. macOS users must update before June 12 or their apps will stop working.
Former OpenAI chief scientist Ilya Sutskever defended his role in Sam Altman's ouster during testimony in the Musk v Altman trial, stating he acted to protect the company from destruction.
Security researchers discovered over 5,000 AI-coded apps from Lovable, Replit, Base44, and Netlify exposing sensitive corporate and personal data without authentication or security measures.
Anthropic introduces a managed runtime for Claude agents, handling tool use, memory, and orchestration so enterprise teams can ship agents without managing the runtime themselves.
More than 70 civil liberties organizations demand Meta abandon facial recognition plans for Ray-Ban smart glasses, warning the 'Name Tag' feature would enable stalkers and predators to identify strangers in public.
