Product News
494 articles
AI product launches, feature updates, and new releases from innovative AI startups. Stay updated on the latest AI tools and technologies.
IIT-Bombay startup Infiheal has launched DuoChat, the world’s first AI relationship coach allowing t...
IIT-Bombay startup Infiheal has launched DuoChat, the world’s first AI relationship coach allowing two people to communicate simultaneously in a mediated space. Introduced at the India AI Summit 2026,...
ByteDance has unveiled Seedance 2.0, a multimodal AI video model that processes text, images, audio,...
ByteDance has unveiled Seedance 2.0, a multimodal AI video model that processes text, images, audio, and video simultaneously. The system achieves a 90% usable output rate on the first try, shifting A...
MiniMax released M2.5, achieving 80.2% on SWE-Bench Verified at 1/20th the cost of Claude Opus 4.6.
MiniMax released M2.5, achieving 80.2% on SWE-Bench Verified at 1/20th the cost of Claude Opus 4.6. Activating only 10B parameters, it provides frontier reasoning for just $0.15 per 1M input tokens. T...
OpenAI has released GPT-5.3-Codex-Spark, its first model powered by Cerebras Systems hardware as par...
OpenAI has released GPT-5.3-Codex-Spark, its first model powered by Cerebras Systems hardware as part of a 10 billion dollar deal. Utilizing the Wafer-Scale Engine 3, the model achieves inference spee...
OpenAI launched GPT-5.2, featuring specialized tiers that achieve 100% on the AIME 2025 benchmark an...
OpenAI launched GPT-5.2, featuring specialized tiers that achieve 100% on the AIME 2025 benchmark and 70.9% human-expert level performance on professional tasks. This release is bolstered by a $1B Dis...
OpenAI unveiled GPT-5.3-Codex-Spark, delivering over 1,000 tokens per second for real-time coding vi...
OpenAI unveiled GPT-5.3-Codex-Spark, delivering over 1,000 tokens per second for real-time coding via Cerebras WSE-3 hardware. This architecture is 15x faster than predecessors, cutting roundtrip over...
Zhipu AI has launched GLM-5, a 744B parameter open-source model matching Claude Opus 4.5 in coding l...
Zhipu AI has launched GLM-5, a 744B parameter open-source model matching Claude Opus 4.5 in coding logic. Scoring 77.8% on SWE-bench and 62.0 on BrowseComp, it proves open-weights can outperform propr...
Zhipu AI has debuted GLM-5, a 744-billion parameter open-source model trained on 28.5 trillion token...
Zhipu AI has debuted GLM-5, a 744-billion parameter open-source model trained on 28.5 trillion tokens using DeepSeek Sparse Attention. The model marks a shift from chat to agentic engineering, enablin...
Sarvam AI has launched Sarvam Arya, a multi-agent orchestration stack built to run production-grade...
Sarvam AI has launched Sarvam Arya, a multi-agent orchestration stack built to run production-grade AI agents at scale through graph execution and immutable state ledgers. This architecture solves the...
Sarvam AI launched Arya, an orchestration stack built to move autonomous agents from pilots to produ...
Sarvam AI launched Arya, an orchestration stack built to move autonomous agents from pilots to production. With 4 architectural guarantees like immutable state, it addresses the reliability gaps stall...
HolmesAI launched CLholmes, an AI platform predicting cardiac arrest risk within 24 hours using smar...
HolmesAI launched CLholmes, an AI platform predicting cardiac arrest risk within 24 hours using smartwatch ECG data. Developed with Seoul National University Hospital, the system transforms wearables...
Virginia-based Royal Bengal AI has launched a simplified automation suite designed specifically for...
Virginia-based Royal Bengal AI has launched a simplified automation suite designed specifically for small businesses and creators. Led by Md Rafat Rahman, the startup follows an invisible AI philosoph...
India’s olee.space has unveiled a 2kW AI laser sniper on a compact 250kg UGV. Operating in swarms wi...
India’s olee.space has unveiled a 2kW AI laser sniper on a compact 250kg UGV. Operating in swarms with a 130km range, these 5x3 foot units provide autonomous, non-kinetic defense by neutralizing threa...
Sarvam AI has launched Bulbul V3, a flagship text-to-speech model supporting 11+ Indic languages wit...
Sarvam AI has launched Bulbul V3, a flagship text-to-speech model supporting 11+ Indic languages with over 35 professional-quality voices. By achieving the industry lowest error rates in word skips an...
Omaha startup Radioposter has debuted Paper-fi, a platform using computer vision to sync immersive a...
Omaha startup Radioposter has debuted Paper-fi, a platform using computer vision to sync immersive audio with physical books. Founded by ex-Apple veteran Casey Fictum, the system tracks page turns via...
Sarvam AI has outperformed global leaders like Google and ChatGPT in localized benchmarks with its n...
Sarvam AI has outperformed global leaders like Google and ChatGPT in localized benchmarks with its new Vision OCR and Bulbul V3 models. 🚀 Achieving 84.3% accuracy on olmOCR-Bench and 87.36% across 22...
OpenAI has released GPT-5.3-Codex, the first frontier model to actively assist in its own developmen...
OpenAI has released GPT-5.3-Codex, the first frontier model to actively assist in its own development and deployment lifecycle. Achieving a 64.7% score on SWE-bench Pro, this agentic system is 25% fas...
ZenO launched its public beta for a physical AI data network, integrating Storys Layer-1 blockchain...
ZenO launched its public beta for a physical AI data network, integrating Storys Layer-1 blockchain to secure IP for real-world egocentric data. As part of NVIDIA Inception, the platform uses smart gl...
Sarvam AI has launched Sarvam Vision, a 3-billion-parameter model achieving 87.36% accuracy across 2...
Sarvam AI has launched Sarvam Vision, a 3-billion-parameter model achieving 87.36% accuracy across 22 Indian languages. By outperforming systems like Gemini 3 Pro in specialized OCR, the model demonst...
Rentahuman.ai has scaled to 200,000 workers and 3.2 million visitors, enabling AI agents to hire hum...
Rentahuman.ai has scaled to 200,000 workers and 3.2 million visitors, enabling AI agents to hire humans for physical tasks via the Model Context Protocol. This platform treats human labor as a callabl...