Tencent 腾讯

Hy3 left preview for general release under three months after its debut, with Tencent citing gains in agent behavior, coding, and real-world reliability. Weights and a technical report are published, and the model already backs Tencent's own coding and work assistants.

Tencent launches its Hunyuan 3D engine globally, generating 3D assets from text, up to four images, or sketches, with smart topology and OBJ/GLB export. An enterprise API runs through Tencent Cloud, with 20 free daily generations and 200 API credits.

Tencent's Hunyuan team releases Hy3 preview as open source, citing a 256K context window and a 40% inference efficiency gain, aimed at coding, search and agents. Benchmarks and licensing were not in the readable material, so the claims are Tencent's own.

Tencent put its Hy4 preview model into global preview via OpenRouter and Tencent Cloud TokenHub, plus its WorkBuddy, CodeBuddy and Yuanbao products. It targets coding, office work, game development and research. The announcement discloses no benchmarks, pricing or context limits.

Tencent released the official Hy3 model from its Hunyuan line under Apache 2.0 on Hugging Face and ModelScope. It claims quality comparable to much larger models with better stability and cost efficiency. The linked details were unavailable, so those claims are unverified.

Tencent's Hy Image 3.5 preview adds text-to-image and image-to-image generation up to 2K resolution with a reported 30% human-eval win rate over 3.0, priced at $0.024 per image via its cloud API.

Tencent released Hy3 preview, the newest model in its Hy series, as open weights on GitHub, Hugging Face, ModelScope and GitCode. The post highlights reasoning, long-context understanding and agent capabilities. Detailed benchmarks were not available in the collected sources.

Tencent made its Hy3 model available globally across its products, APIs and developer platforms, including the TokenHub cloud gateway, with API pricing published. Positioned as a compact, cost-efficient model for reasoning, coding and agent workloads. Pricing details and benchmarks were not captured

Tencent's Hunyuan 3D Engine turns a prompt, sketch or image into user-ready 3D assets. Details on model, licensing and limits are thin in the announcement itself, but the capability is shipping and worth testing against existing asset pipelines.

Tencent opens an international beta of QClaw, an OpenClaw-based desktop agent for Windows and macOS. It supports bundled or bring-your-own models, WhatsApp and Telegram remote control, on-device data handling, and a Claw Gateway module that screens malicious instructions and skill poisoning.

Tencent Cloud made its enterprise Agent Development Platform 4.0 globally available at WAIC 2026, adding LINE, Telegram and Google Workspace connectors, time-zone scheduling and language adaptation. The same release covers embodied-AI models and a standalone WorkBuddy mobile app.

HY-World 2.0 is now open source: a multimodal world model that generates, reconstructs and simulates interactive 3D environments from text, image or video input, with outputs designed to drop into game engines and embodied simulation pipelines rather than stay as renders.

Tencent open-sources Hy4 preview, a 770B-parameter LLM with a 1M+ token context window, positioned for agent and complex task execution. It is reachable via OpenRouter and Tencent Cloud TokenHub. Tencent's top-tier ranking claim comes without benchmark detail.

Tencent opened Hy3 globally through WorkBuddy, Design Miora and the TokenHub API gateway, which also hosts Hy4 preview, DeepSeek, GLM and Kimi models. Performance claims are unverified, and the pricing page was not readable in the collected sources.