Llm
- All
- News
- Videos
- Web Stories
-
SenseTime's Open-Source SenseNova-SI AI Model Said to Outperform GPT-5, Gemini 2.5 Pro in Spatial Intelligence
SenseTime has released two open-source AI models, SenseNova-SI-InternVL3-2B and 8B, claiming they outperform OpenAI’s GPT-5 and Google’s Gemini 2.5 Pro in spatial intelligence. The models reportedly excel at understanding 3D object relationships and motion, crucial for robotics and mapping. Available on Hugging Face under Apache 2.0 license, de...
-
Grab Superapp Says AI Models Struggle to Understand Asian Languages
Grab says it built its own lightweight vision model after finding both proprietary and open-source AI systems struggled with Southeast Asian languages, leading to errors and high latency in document processing. The move underscores a broader concern: frontier models from firms like Google, OpenAI and Anthropic still lack robust multilingual support...
-
Apple CEO Reportedly Confirms Partnership Plans Beyond OpenAI; Revamped Siri Expected to Launch in 2026
Apple has intentions of integrating with more artificial intelligence (AI) companies beyond OpenAI, CEO Tim Cook said in a conversation with a publication on Thursday. The Cupertino-based tech giant has long been rumoured to form partnerships with AI firms to have their large language models (LLMs) power features on iPhone and other Apple devices. ...
-
Carnegie Mellon’s AI Drones Can Build Mid-Air Structures With 90 Percent Success Rate
Carnegie Mellon engineers have unveiled AI-controlled drones that 3D print by snapping magnetic blocks together mid-air. Powered by a large language model, the drones translate simple commands into precise construction plans, achieving 90% accuracy. The technology could enable aerial infrastructure repair and emergency shelter construction in hard-...
-
"India Working On 2 nm Chip": Ashwini Vaishnaw Shows 'Wafer' At NDTV Summit
Digital credit, fast mobile data and large language models (LLMs) - these are among some of the initiatives that India has taken in its digital-led growth story, Union Minister Ashwini Vaishnaw said at the NDTV World Summit in Delhi today.
-
Anthropic Releases Claude Haiku 4.5 as a Fast and Cost-Effective AI Model
Anthropic has launched Claude Haiku 4.5, the smallest model in its Claude 4.5 series. Focused on speed and real-time performance, it’s said to be twice as fast as Sonnet 4 while maintaining similar coding capabilities. Available for free, Haiku 4.5 outperforms rival models in several benchmarks and is rated Anthropic’s safest model to date.
-
Anthropic Warns That Minimal Data Contamination Can ‘Poison’ Large AI Models
Anthropic has warned that even a few poisoned samples in a dataset can compromise an AI model. A joint study with the UK AI Security Institute found that as few as 250 malicious documents can implant backdoors in LLMs up to 13B parameters, proving model size offers no protection.
-
MeitY's IndiaAI Mission Taps Five Projects to Drive Safe, Trusted AI in the Country
IndiaAI Mission, under the Ministry of Electronics and Information Technology (MeitY), has selected five artificial intelligence (AI) projects aimed at making AI systems safer and fairer. Announced on Monday, these projects emerged from over 400 proposals received under the “Safe & Trusted AI” pillar launched in December 2024. The selected init...
-
Does AI Pose An Existential Risk? What Experts Say
There are many claims to sort through in the current era of ubiquitous artificial intelligence (AI) products, especially generative AI ones based on large language models or LLMs, such as ChatGPT, Copilot, Gemini and many, many others.
-
What Is Dead Internet Theory? ChatGPT Boss Sam Altman Hints At Truth Behind Conspiracy
OpenAI CEO Sam Altman hinted that the dead internet theory may not be a conspiracy anymore, owing to the rise of bot accounts on X.
-
Microsoft Unveils Its First Fully Homegrown AI Models; Set to Bring New Features to Copilot
Microsoft unveiled the company's first two homegrown artificial intelligence (AI) models on Thursday. The foundation models were developed by the Redmond-based tech giant’s AI division, and is said to be built entirely in-house. The first model is MAI-Voice-1, a speech generation model that natively generates expressive and natural sounding voice...
-
Resurrecting the Dead Using AI: What Are Deadbots and Why Are They Stirring Ethical Debate?
Deadbots are trained on the digital footprints of an individual, such as text messages, emails, voice recordings, videos, social media profiles, and more. A large language model (LLM) is trained on this dataset, and it can generate responses that appear similar to what the person would have said, in the manner that they would have said.
-
Beatoven.ai Unveils Maestro AI Music Generation Model, Announces Revenue-Sharing for Artists
Beatoven.ai, a Bengaluru-based artificial intelligence (AI) startup focused on music generation, unveiled its first in-house AI Model on Wednesday. Dubbed Maestro, it is a music model that can generate instrumental tracks across various genres based on text prompts. The company claims that the large language model (LLM) was only trained on fully li...
-
OpenAI Says GPT-5 Is Now ‘Warmer and Friendlier’ After Users Complained About Cold Treatment
OpenAI updated the GPT-5 artificial intelligence (AI) model on Saturday to make its responses “warmer and friendlier.” The San Francisco-based AI firm highlighted that it has tweaked the model so that it would not appear cold or rude to users. The updated version of the large language model (LLM) is currently being rolled out to all users globa...
-
ChatGPT 5 Is Here: What's New, How to Use It, And Is It Really Free?
ChatGPT 5 has officially launched, offering smarter, faster, and more human-like AI interactions. Here's a breakdown of its key new features, how to access and use it, and whether it's truly free for all users.
-
SenseTime's Open-Source SenseNova-SI AI Model Said to Outperform GPT-5, Gemini 2.5 Pro in Spatial Intelligence
SenseTime has released two open-source AI models, SenseNova-SI-InternVL3-2B and 8B, claiming they outperform OpenAI’s GPT-5 and Google’s Gemini 2.5 Pro in spatial intelligence. The models reportedly excel at understanding 3D object relationships and motion, crucial for robotics and mapping. Available on Hugging Face under Apache 2.0 license, de...
-
Grab Superapp Says AI Models Struggle to Understand Asian Languages
Grab says it built its own lightweight vision model after finding both proprietary and open-source AI systems struggled with Southeast Asian languages, leading to errors and high latency in document processing. The move underscores a broader concern: frontier models from firms like Google, OpenAI and Anthropic still lack robust multilingual support...
-
Apple CEO Reportedly Confirms Partnership Plans Beyond OpenAI; Revamped Siri Expected to Launch in 2026
Apple has intentions of integrating with more artificial intelligence (AI) companies beyond OpenAI, CEO Tim Cook said in a conversation with a publication on Thursday. The Cupertino-based tech giant has long been rumoured to form partnerships with AI firms to have their large language models (LLMs) power features on iPhone and other Apple devices. ...
-
Carnegie Mellon’s AI Drones Can Build Mid-Air Structures With 90 Percent Success Rate
Carnegie Mellon engineers have unveiled AI-controlled drones that 3D print by snapping magnetic blocks together mid-air. Powered by a large language model, the drones translate simple commands into precise construction plans, achieving 90% accuracy. The technology could enable aerial infrastructure repair and emergency shelter construction in hard-...
-
"India Working On 2 nm Chip": Ashwini Vaishnaw Shows 'Wafer' At NDTV Summit
Digital credit, fast mobile data and large language models (LLMs) - these are among some of the initiatives that India has taken in its digital-led growth story, Union Minister Ashwini Vaishnaw said at the NDTV World Summit in Delhi today.
-
Anthropic Releases Claude Haiku 4.5 as a Fast and Cost-Effective AI Model
Anthropic has launched Claude Haiku 4.5, the smallest model in its Claude 4.5 series. Focused on speed and real-time performance, it’s said to be twice as fast as Sonnet 4 while maintaining similar coding capabilities. Available for free, Haiku 4.5 outperforms rival models in several benchmarks and is rated Anthropic’s safest model to date.
-
Anthropic Warns That Minimal Data Contamination Can ‘Poison’ Large AI Models
Anthropic has warned that even a few poisoned samples in a dataset can compromise an AI model. A joint study with the UK AI Security Institute found that as few as 250 malicious documents can implant backdoors in LLMs up to 13B parameters, proving model size offers no protection.
-
MeitY's IndiaAI Mission Taps Five Projects to Drive Safe, Trusted AI in the Country
IndiaAI Mission, under the Ministry of Electronics and Information Technology (MeitY), has selected five artificial intelligence (AI) projects aimed at making AI systems safer and fairer. Announced on Monday, these projects emerged from over 400 proposals received under the “Safe & Trusted AI” pillar launched in December 2024. The selected init...
-
Does AI Pose An Existential Risk? What Experts Say
There are many claims to sort through in the current era of ubiquitous artificial intelligence (AI) products, especially generative AI ones based on large language models or LLMs, such as ChatGPT, Copilot, Gemini and many, many others.
-
What Is Dead Internet Theory? ChatGPT Boss Sam Altman Hints At Truth Behind Conspiracy
OpenAI CEO Sam Altman hinted that the dead internet theory may not be a conspiracy anymore, owing to the rise of bot accounts on X.
-
Microsoft Unveils Its First Fully Homegrown AI Models; Set to Bring New Features to Copilot
Microsoft unveiled the company's first two homegrown artificial intelligence (AI) models on Thursday. The foundation models were developed by the Redmond-based tech giant’s AI division, and is said to be built entirely in-house. The first model is MAI-Voice-1, a speech generation model that natively generates expressive and natural sounding voice...
-
Resurrecting the Dead Using AI: What Are Deadbots and Why Are They Stirring Ethical Debate?
Deadbots are trained on the digital footprints of an individual, such as text messages, emails, voice recordings, videos, social media profiles, and more. A large language model (LLM) is trained on this dataset, and it can generate responses that appear similar to what the person would have said, in the manner that they would have said.
-
Beatoven.ai Unveils Maestro AI Music Generation Model, Announces Revenue-Sharing for Artists
Beatoven.ai, a Bengaluru-based artificial intelligence (AI) startup focused on music generation, unveiled its first in-house AI Model on Wednesday. Dubbed Maestro, it is a music model that can generate instrumental tracks across various genres based on text prompts. The company claims that the large language model (LLM) was only trained on fully li...
-
OpenAI Says GPT-5 Is Now ‘Warmer and Friendlier’ After Users Complained About Cold Treatment
OpenAI updated the GPT-5 artificial intelligence (AI) model on Saturday to make its responses “warmer and friendlier.” The San Francisco-based AI firm highlighted that it has tweaked the model so that it would not appear cold or rude to users. The updated version of the large language model (LLM) is currently being rolled out to all users globa...
-
ChatGPT 5 Is Here: What's New, How to Use It, And Is It Really Free?
ChatGPT 5 has officially launched, offering smarter, faster, and more human-like AI interactions. Here's a breakdown of its key new features, how to access and use it, and whether it's truly free for all users.
