# open source
Latest news and articles about open source
Total: 12 articles found
Alibaba’s Qwen3.5 Claims Gemini‑3‑Pro Parity at a Fraction of the Cost — A Shift from Scale to Efficiency
Alibaba has open‑sourced Qwen3.5‑Plus, a 397B‑parameter multimodal model the company says matches Gemini 3 Pro’s performance while operating with only ~17B activated parameters and much lower inference costs. The model emphasises architectural efficiency, native multimodal pretraining and agent capabilities, and forms part of a flurry of Chinese model launches that shift competition from raw scale to systems and cost efficiency.
OpenAI Poaches OpenClaw Founder as It Places a Big Bet on Autonomous AI Agents
OpenAI has hired Peter Steinberger, founder of the open‑source AI agent project OpenClaw, and will place OpenClaw under a foundation supported by OpenAI. The move underscores OpenAI’s bet that multi‑agent, action‑oriented systems are the next major product frontier, while rekindling concerns about whether corporate sponsorship will erode open‑source independence.
OpenAI Recruits Creator of OpenClaw, Vows to Keep Viral Agent Open-Source via New Foundation
OpenAI has hired Peter Steinberger, creator of the widely adopted agent framework OpenClaw, and pledged to place the project into a foundation that will keep it open-source and independent while receiving funding and support. The move is a tactical win for OpenAI but raises questions about governance, security and the balance between openness and centralization as agent platforms mature.
Huawei’s Ascend Hardware Quickly Hosts Open‑Source MiniMax M2.5, Underscoring China’s Push for a Homegrown AI Stack
Xiyu Technology’s open‑source release of MiniMax M2.5 was ported within hours to Huawei’s Ascend‑based Atlas 800 A2 and A3 servers and trialed at operational network sites. The episode highlights accelerating co‑design between Chinese model developers and domestic hardware vendors, a trend that strengthens China’s push for an autonomous AI infrastructure.
China’s Zhipu AI Soars After GLM‑5 Open‑Source Push — But Profits Remain Distant
Zhipu AI’s shares surged after open‑sourcing its GLM‑5 foundation model and hiking prices for an AI coding subscription, boosting investor optimism about product traction. Yet the company continues to incur large losses driven by heavy R&D and compute costs, and is pursuing a mainland Shanghai listing to broaden its capital base.
Seedance 2.0: China’s Groundbreaking AI Video Engine That Both Liberates and Alarm
Seedance 2.0, a Chinese generative video model, can produce hyper-realistic footage by integrating image, motion, audio and text, lowering the barriers to making sophisticated video. Its realism has sparked copyright backlash and deepfake concerns even as China’s large user base, open-source releases and supportive AI policy accelerate development and adoption.
China’s Haiguang DCU Enables Day‑One Integration of Zhipu AI’s Open‑Source GLM‑5
Zhipu AI released GLM‑5 as open source on Feb. 11, and Haiguang DCU completed Day‑0 adaptation and joint fine‑tuning to provide immediate, deployable solutions for developers and enterprises. The move highlights accelerating integration between Chinese model creators and infrastructure providers, shortening the path from research release to commercial use.
OpenClaw’s Viral Rise Signals a New Age for Cheap, Deployable AI Agents — and New Risks
OpenClaw, an open‑source agent platform created by Peter Steinberger, has gone viral by turning chat messages into executable commands across multiple model APIs, accelerating demand for inexpensive, high‑throughput models and simple local hardware like the Mac Mini. The surge highlights opportunities for Chinese model providers such as Minimax and Kimi, while raising acute security, deployment and governance challenges.
SenseTime Open-Sources ‘Sense Nova‑MARS,’ Betting on Agentic Multimodal AI to Drive Execution‑Capable Applications
SenseTime has open‑sourced Sense Nova‑MARS, a multimodal Agentic VLM available in 8B and 32B parameter sizes that the company says can plan actions, call tools and deeply fuse dynamic visual reasoning with image‑text search. The move democratizes access to execution‑oriented multimodal models, accelerating research and product integration while raising safety and governance questions about agentic AI.
China’s 2026 AI Crossroads: Open‑source Infrastructure as the Antidote to Vendor Lock‑in
SUSE executives warn that 2026 is a decisive year for Chinese firms adopting AI: they must choose between the short‑term convenience of closed platforms and the long‑term freedom of open, interoperable infrastructure. SUSE positions SLES 16, long support cycles and observability tooling as solutions to vendor lock‑in, regulatory demands and runaway compute costs.
Alibaba Open‑Sources Qwen3‑TTS, Bringing Multilingual Voice Cloning to Developers
Alibaba’s Qwen team has open‑sourced Qwen3‑TTS, a family of text‑to‑speech models in 1.7B and 0.6B sizes supporting voice cloning and multilingual, dialect‑aware synthesis. The release aims to broaden developer access and accelerate voice applications while raising urgent questions about misuse, detection and governance.
Small, Open and Multimodal: Chinese Startup Releases 10‑Billion‑Parameter Vision‑Language Model Claiming SOTA Performance
Chinese start‑up Jieyue Xingchen open‑sourced Step3‑VL‑10B, a 10‑billion‑parameter multimodal model that the team says matches same‑scale state‑of‑the‑art performance on vision, reasoning, math and dialogue. The release highlights a push for efficient, deployable multimodal models and will prompt independent verification and community adoption.