Model Launch

Twitter

Facebook

Copy Link

Top News

Guide Labs open-sources Steerling-8B, a new interpretable LLM that traces every token to source

San Francisco startup Guide Labs has open-sourced Steerling-8B, an 8-billion parameter language model built with a novel architecture that makes every generated token traceable to its training data origins. Founded by Julius Adebayo and Aya Abdelsalam Ismail, the company aims to solve AI's black box problem by engineering interpretability directly into the model rather than analyzing it post-hoc.

Technology2 sources

Guide Labs open-sources Steerling-8B, a new interpretable LLM that traces every token to source

Technology2 sources

Cybersecurity Stocks Tumble on AI Threat Concerns as Analysts Call Investor Fears Overdone

CrowdStrike and Palo Alto Networks experienced steep declines following Anthropic's launch of Claude Code, an AI-powered security tool. But industry experts argue the sell-off in cybersecurity stocks reflects panic rather than fundamentals, with AI actually creating the biggest opportunity in cybersecurity history as threats accelerate and attack surfaces expand.

Business2 sources

Cybersecurity Stocks Tumble on AI Threat Concerns as Analysts Call Investor Fears Overdone

Business2 sources

Google Gemini 3.1 Pro doubles reasoning score, beats rivals in key AI benchmarks

Google released Gemini 3.1 Pro with dramatically improved problem-solving and reasoning capabilities, more than doubling its predecessor's performance on abstract reasoning tests. The newest and more powerful large language model scored 77.1% on ARC-AGI-2 and achieved a record 44.4% on Humanity's Last Exam, outperforming OpenAI and Anthropic in most benchmarks as accelerating competition among major tech companies intensifies.

Technology27 sources

Google Gemini 3.1 Pro doubles reasoning score, beats rivals in key AI benchmarks

Technology27 sources

India unveils homegrown AI models at major summit, but experts say DeepSeek moment remains distant

Indian AI startups showcased locally-built models at New Delhi's AI Impact Summit, with Prime Minister Modi celebrating 'Made in India' innovation. Sarvam AI released models optimized for 22 Indian languages, while Gnani.ai launched Vachana speech models trained on over a million hours of audio. But analysts suggest India's realistic path lies in becoming the world's largest AI adoption market rather than a frontier innovation hub.

Technology2 sources

India unveils homegrown AI models at major summit, but experts say DeepSeek moment remains distant

Technology2 sources

Google Gemini rolls out Lyria 3 to generate 30-second AI songs from text and images

Google has integrated DeepMind's Lyria 3 model into the Gemini app, enabling users worldwide to create 30-second music tracks from simple text or image prompts. The feature auto-generates lyrics and produces custom cover art, marking a significant expansion of AI music generation capabilities. While Google emphasizes copyright protections and SynthID watermarking, the move raises questions about artistic authenticity and the music industry's future.

Technology20 sources

Google Gemini rolls out Lyria 3 to generate 30-second AI songs from text and images

Technology20 sources

Anthropic releases Claude Sonnet 4.6 with human-level computer use and coding improvements

Anthropic launched Claude Sonnet 4.6, bringing frontier-level AI capabilities to free and Pro users. The update delivers improved coding skills that rival its premium Opus model, human-level performance in computer use tasks, and a 1 million token context window. Developers prefer the new model 70% of the time over its predecessor, with major gains in security against prompt injection attacks.

Technology18 sources

Anthropic releases Claude Sonnet 4.6 with human-level computer use and coding improvements

Technology18 sources

Gnani.ai unveils Inya VoiceOS, India's first 5-billion-parameter voice-to-voice AI model

Indian AI startup Gnani.ai launched Inya VoiceOS at the India AI Impact Summit 2026, marking a significant milestone as India's first voice-to-voice AI system. The 5-billion-parameter model processes speech directly without converting to text, trained on 14 million hours of multilingual Indian speech data. Prime Minister Narendra Modi formally unveiled the system, which supports over 15 Indian languages and handles code-mixed conversations.

Technology2 sources

Gnani.ai unveils Inya VoiceOS, India's first 5-billion-parameter voice-to-voice AI model

Technology2 sources

Google AI upgrades Gemini 3 Deep Think to tackle advanced scientific research challenges

Google has released a major upgrade to Gemini 3 Deep Think, its specialized AI reasoning model designed for scientific research. The update achieves record-breaking scores on benchmarks like ARC-AGI-2 (84.6%) and Humanity's Last Exam (48.4%), outperforming OpenAI's GPT-5.2 and Anthropic's Claude Opus 4.6. Real-world applications include identifying peer review flaws in mathematics papers and optimizing semiconductor fabrication methods.

Technology4 sources

Google AI upgrades Gemini 3 Deep Think to tackle advanced scientific research challenges

Technology4 sources

Microsoft pushes for AI self-sufficiency as it builds in-house models to reduce OpenAI dependence

Microsoft's AI chief Mustafa Suleyman announced the company is pursuing AI self-sufficiency by developing its own advanced foundation models. The strategic shift aims to reduce reliance on OpenAI while maintaining their partnership through 2032. Microsoft previewed MAI-1, trained on 15,000 NVIDIA H100 GPUs, and launched the Maia 200 inference chip to control AI economics.

Business3 sources

Microsoft pushes for AI self-sufficiency as it builds in-house models to reduce OpenAI dependence

Business3 sources

Zhipu Unveils GLM-5 Model With 744B Parameters, Intensifying Chinese AI Race With DeepSeek

Chinese AI startup Zhipu released GLM-5, a new flagship model with 744 billion parameters that rivals Anthropic's Claude Opus in coding benchmarks. Trained entirely on Huawei Ascend chips, the open-source model achieved a record-low hallucination rate and sent Zhipu shares surging 34%. The release intensifies competition among Chinese AI companies ahead of DeepSeek's anticipated next-generation model launch.

Technology5 sources

Zhipu Unveils GLM-5 Model With 744B Parameters, Intensifying Chinese AI Race With DeepSeek

Technology5 sources

ChatGPT Deep Research now runs on GPT-5.2 with source controls and real-time tracking

OpenAI has upgraded its Deep Research feature in ChatGPT with the GPT-5.2 model, introducing the ability to specify research sources, a full-screen report viewer with citations, and real-time progress tracking. The upgrades to Deep Research feature aim to deliver more accurate, citation-backed insights for professionals conducting complex research tasks.

Technology3 sources

ChatGPT Deep Research now runs on GPT-5.2 with source controls and real-time tracking

Technology3 sources

OpenAI faces California AI safety law violations over high-risk GPT-5.3-Codex release

OpenAI is under fire from watchdog group The Midas Project for allegedly violating California's new AI safety law with its GPT-5.3-Codex coding model release. The company's first model to hit "high" cybersecurity risk failed to implement legally required safeguards, raising questions about compliance with the state's SB 53 regulations that came into effect in January.

Policy2 sources

OpenAI faces California AI safety law violations over high-risk GPT-5.3-Codex release

Policy2 sources

Blackstone boosts stake in Anthropic to $1 billion as AI startup hits $350 billion valuation

Blackstone is increasing its investment in AI startup Anthropic by $200 million, bringing its total stake to approximately $1 billion at a $350 billion valuation. The move reflects surging investor interest in AI as Anthropic's funding round exceeds its initial $10 billion target by more than double, driven by strong demand for generative AI companies developing advanced large language models.

Startups2 sources

Blackstone boosts stake in Anthropic to $1 billion as AI startup hits $350 billion valuation

Startups2 sources

Chile unveils Latam-GPT, an open-source AI model built to combat bias in Latin America

Chile launched Latam-GPT, the first open-source artificial intelligence language model trained on Latin American cultures. Developed by CENIA with $550,000 in funding, the model uses over eight terabytes of regional data to address linguistic biases in AI systems trained primarily on English content. The initiative positions Latin America as an active player in the global AI race.

Technology4 sources

Chile unveils Latam-GPT, an open-source AI model built to combat bias in Latin America

Technology4 sources

Runway raises $315M at $5.3B valuation as AI video startup pivots to world model development

Runway AI Inc. secured a $315 million Series E funding round led by General Atlantic, nearly doubling its valuation to $5.3 billion. The AI video generation startup plans to invest in world models—AI systems that simulate environments for planning future events—targeting applications in medicine, climate, energy, and robotics beyond its traditional media and entertainment base.

Startups4 sources

Runway raises $315M at $5.3B valuation as AI video startup pivots to world model development

Startups4 sources

Prev 1…7 8910 Next