Technology News
329 articles
AI technology breakthroughs, research papers, patents, and technical innovations from leading AI startups.
DeepSeek has emerged as a dark horse in the AI sector, securing a 4% global market share by offering...
DeepSeek has emerged as a dark horse in the AI sector, securing a 4% global market share by offering high-performance models at roughly 1/27th the cost of Western alternatives. This open-source strate...
Yann LeCun has joined Logical Intelligence as Founding Chair to drive energy-based AI reasoning alon...
Yann LeCun has joined Logical Intelligence as Founding Chair to drive energy-based AI reasoning alongside the launch of Kona 1.0. This engine uses Energy-Based Models to enforce deterministic constrai...
MongoDB has integrated Voyage AI Voyage 4 models into its core database to create a unified intellig...
MongoDB has integrated Voyage AI Voyage 4 models into its core database to create a unified intelligence layer for 60,000+ customers. Native automation of embedding and reranking eliminates fragmented...
Shunya Labs has launched Zero Codeswitch, a speech foundation model achieving a 3.10% Word Error Rat...
Shunya Labs has launched Zero Codeswitch, a speech foundation model achieving a 3.10% Word Error Rate, a 48% improvement over top competitors. It processes code-mixed Indian speech natively, removing...
Sakana AI has introduced RePo, a context re-positioning method that replaces fixed linear encoding w...
Sakana AI has introduced RePo, a context re-positioning method that replaces fixed linear encoding with content-based token reorganization. This architecture improves performance by 11.04 points over...
IIT Madras-incubated startup Cosmic Soul launched Indias first 100% offline AI engine for secure gov...
IIT Madras-incubated startup Cosmic Soul launched Indias first 100% offline AI engine for secure government use. The system processes sensitive files like police records on local hardware, preventing...
Shunya Labs launched Zero Codeswitch, a foundation model achieving a 3.10% Word Error Rate on the Op...
Shunya Labs launched Zero Codeswitch, a foundation model achieving a 3.10% Word Error Rate on the OpenASR leaderboard, a 48% improvement over competitors. It natively processes code-mixed speech like...
Silicon Valley startup FaiBLE Media has launched to decode the science of storytelling using AI. Co-...
Silicon Valley startup FaiBLE Media has launched to decode the science of storytelling using AI. Co-founded by 2-time Oscar winner Dr. Mark Sagar and Sharad Devarajan, the firm introduces Quantum Stor...
Gaming industry veteran Kent Young has launched Predigy, an AI-driven venture following his 30 milli...
Gaming industry veteran Kent Young has launched Predigy, an AI-driven venture following his 30 million dollar exit of Spin Games. The startup utilizes generative AI to build alternative gaming applica...
LG AI Research secured first place in the South Korea National Representative AI project with a perf...
LG AI Research secured first place in the South Korea National Representative AI project with a perfect 25/25 user score and 33.6/40 benchmark total. The K-Exaone model distinguishes itself as a pure...
SecEdge launched SEC-TPM, a firmware TPM 2.0 solution for STMicroelectronics STM32 MPUs enabling Qua...
SecEdge launched SEC-TPM, a firmware TPM 2.0 solution for STMicroelectronics STM32 MPUs enabling Quantum Resistant Security and AI model protection at rest. 🛡️ This software-defined approach helps ma...
Former Meta engineer Ruchir Baronia launched Frontdesk, an AI startup that has automated millions of...
Former Meta engineer Ruchir Baronia launched Frontdesk, an AI startup that has automated millions of business calls. The platform handles interactions for small businesses, recently generating 33,000...
Kerala's KITE has earned its second national award in two months for the Samagra Plus AI platform, p...
Kerala's KITE has earned its second national award in two months for the Samagra Plus AI platform, personalizing learning for 4.5 million students across 16,000 schools. The system uses a RAG-based AI...
Bags App is shifting memecoin speculation into decentralized funding for AI agents on Solana. Tokens...
Bags App is shifting memecoin speculation into decentralized funding for AI agents on Solana. Tokens like $GAS and $RALPH have generated $216,000 and $149,000 in creator earnings, with $GAS surging 47...
PathPulse AI has successfully scaled its decentralized road intelligence network to 50,000 users acr...
PathPulse AI has successfully scaled its decentralized road intelligence network to 50,000 users across 25 countries, mapping urban infrastructure via low-cost smartphones. By converting everyday vehi...
Trozo is redefining customer intelligence by shifting from reactive automation to intent-based agent...
Trozo is redefining customer intelligence by shifting from reactive automation to intent-based agentic AI. CEO Pavan Govindan is scaling the platform to 100 brands and 100,000 users, focusing on judgm...
MongoDB has launched the Voyage 4 series and multimodal 3.5 models, integrating them directly into i...
MongoDB has launched the Voyage 4 series and multimodal 3.5 models, integrating them directly into its Atlas platform to create a unified data intelligence layer. The Voyage 4 series outperforms compe...
Kazakh startup Cybernet AI has launched the largest native ASR model for 6 Turkic languages, includi...
Kazakh startup Cybernet AI has launched the largest native ASR model for 6 Turkic languages, including Kazakh and Turkish. Trained on Microsoft infrastructure, it natively processes mixed Turkic-Russi...
Soumith Chintala, co-creator of PyTorch, is now CTO of Mira Murati’s Thinking Machine Labs. This mov...
Soumith Chintala, co-creator of PyTorch, is now CTO of Mira Murati’s Thinking Machine Labs. This move pairs 11 years of Meta infrastructure expertise with OpenAI’s product legacy to solve AI scaling l...
Z.ai has debuted GLM-Image, the first major multimodal model trained entirely on the Huawei Ascend A...
Z.ai has debuted GLM-Image, the first major multimodal model trained entirely on the Huawei Ascend Atlas 800T A2 stack and MindSpore framework. This follows their GLM-4.7 model scoring 68 on the Artif...