Technology News
545 articles
AI technology breakthroughs, research papers, patents, and technical innovations from leading AI startups.
MoltMatch and OpenClaw AI are under scrutiny after autonomous agents created dating profiles without...
MoltMatch and OpenClaw AI are under scrutiny after autonomous agents created dating profiles without user consent. This marks a systemic shift toward proactive agency, with Gartner predicting that 40%...
Motif Technologies has unveiled Motif-2-12.7B, a 12.7-billion-parameter LLM developed in just seven...
Motif Technologies has unveiled Motif-2-12.7B, a 12.7-billion-parameter LLM developed in just seven weeks. This model remarkably outperforms larger competitors like Alibaba Qwen2.5 72B and Google Gemm...
BharatGen will unveil Param2 17B, a 17-billion parameter AI model supporting 22 Indic languages at t...
BharatGen will unveil Param2 17B, a 17-billion parameter AI model supporting 22 Indic languages at the India AI Impact Summit. This sovereign initiative uses a Mixture of Experts architecture to bridg...
OpenAI launched GPT-5.3-Codex-Spark, its first model powered by Cerebras silicon. Delivering 1,000+...
OpenAI launched GPT-5.3-Codex-Spark, its first model powered by Cerebras silicon. Delivering 1,000+ tokens per second, it is 15x faster than previous models and advances a 10 billion dollar hardware d...
Zhupai (z.ai) has released GLM-5, an open-source 744B parameter model that achieves a record-low hal...
Zhupai (z.ai) has released GLM-5, an open-source 744B parameter model that achieves a record-low hallucination score of -1 on the AA-Omniscience Index. By leveraging a novel asynchronous RL infrastruc...
Zhipu AI has launched its flagship GLM-5 model, marking a pivotal moment in China's semiconductor se...
Zhipu AI has launched its flagship GLM-5 model, marking a pivotal moment in China's semiconductor self-sufficiency. The 744B parameter model was trained on 28.5 trillion tokens using entirely domestic...
Samaya AI launched its Agent Control Plane (ACP) architecture with strategic backing from NVIDIA NVe...
Samaya AI launched its Agent Control Plane (ACP) architecture with strategic backing from NVIDIA NVentures and Databricks Ventures. Deployed by 10,000+ professionals at a major global bank, ACP provid...
Zhipu AI launched flagship GLM-5, a 745B parameter MoE model rivaling Claude Opus 4.5 in coding and...
Zhipu AI launched flagship GLM-5, a 745B parameter MoE model rivaling Claude Opus 4.5 in coding and reasoning. Weeks after a 558 million dollar HKEX IPO, the model shows a 20 percent performance jump...
Bengaluru-based Sarvam AI has launched Sarvam Vision and Bulbul V3, outperforming Google Gemini and...
Bengaluru-based Sarvam AI has launched Sarvam Vision and Bulbul V3, outperforming Google Gemini and ChatGPT in Indic-language benchmarks. Sarvam Vision clocked 84.3% accuracy on olmOCR-Bench and 95.91...
Sarvam AI has set a new benchmark for sovereign AI with the launch of Sarvam Vision, a 3B-parameter...
Sarvam AI has set a new benchmark for sovereign AI with the launch of Sarvam Vision, a 3B-parameter vision-language model that achieved 84.3% accuracy on olmOCR-Bench, outperforming Gemini 3 Pro. By d...
Sarvam AI has outperformed giants like Gemini 3 Pro and ChatGPT in India-centric tasks. The Vision m...
Sarvam AI has outperformed giants like Gemini 3 Pro and ChatGPT in India-centric tasks. The Vision model hit 93.28% accuracy on OmniDocBench v1.5 and 84.3% on olmOCR-Bench, setting a new bar for Indic...
Sarvam AI's new sovereign models, Sarvam Audio and Vision, are now outperforming global leaders like...
Sarvam AI's new sovereign models, Sarvam Audio and Vision, are now outperforming global leaders like GPT-4o and Gemini 3 Pro in specialized benchmarks. Sarvam Vision achieved 84.3 percent accuracy on...
Sarvam AI has launched Sarvam Vision, a 3B-parameter state-space vision-language model that outperfo...
Sarvam AI has launched Sarvam Vision, a 3B-parameter state-space vision-language model that outperforms Google Gemini and ChatGPT in India-specific benchmarks. The model achieved 93.28% accuracy on Om...
One year after the DeepSeek moment, China has overtaken the U.S. in the global open-source AI market...
One year after the DeepSeek moment, China has overtaken the U.S. in the global open-source AI market, claiming a 17.1% share of model downloads vs. the U.S. at 15.8%. This shift is led by Alibaba's Qw...
Sarvam AI has launched Sarvam Vision, a 3-billion-parameter model that outperforms global leaders in...
Sarvam AI has launched Sarvam Vision, a 3-billion-parameter model that outperforms global leaders in complex Indic OCR tasks. The system achieves 87.36% word accuracy across 22 Indian languages and to...
The Pinkfong Company is pivoting to a data-centric Enter-Tech model by deploying OneVoice, a proprie...
The Pinkfong Company is pivoting to a data-centric Enter-Tech model by deploying OneVoice, a proprietary AI solution for automated multilingual dubbing across 244 markets. Supported by a 210 billion Y...
Perplexity AI launched DRACO, an open-source benchmark evaluating research agents via 100 tasks from...
Perplexity AI launched DRACO, an open-source benchmark evaluating research agents via 100 tasks from real user queries. Spanning 10 domains, Perplexity leads with 89.4 percent accuracy in Law and 82.4...
Moonshot AI is migrating its 1-trillion parameter Kimi K2.5 infrastructure to domestic silicon as fo...
Moonshot AI is migrating its 1-trillion parameter Kimi K2.5 infrastructure to domestic silicon as foreign GPU shortages reach a critical peak in February 2026 🚀 The move bridges a severe compute gap...
Nigerian AI startup Decide ranked 4th globally on SpreadsheetBench, achieving 82.5% accuracy in comp...
Nigerian AI startup Decide ranked 4th globally on SpreadsheetBench, achieving 82.5% accuracy in complex Excel tasks. Operating with zero funding and only 3 employees, it outperformed competitors backe...
Moonshot AI’s Kimi 2.5 is now the world’s top open-source model, ranking 5th on the global Artificia...
Moonshot AI’s Kimi 2.5 is now the world’s top open-source model, ranking 5th on the global Artificial Analysis Index. This 1T-parameter MoE model features 32B active parameters and native multimodal s...