Microsoft Phi-4 Rivals GPT-4 on Reasoning
Microsoft's Phi-4 small language model matches GPT-4 performance on key reasoning benchmarks while running on a fraction…
📂 LLM News
Latest articles in LLM News
OpenAI GPT-5 Turbo Debuts With Multimodal Reasoning
OpenAI launches GPT-5 Turbo featuring native multimodal reasoning across text, image, audio, and video inputs in a singl…
Anthropic Doubles Claude Rate Limits, Kills Peak Throttling
Anthropic doubles 5-hour usage quotas for all paid Claude subscribers and removes peak-hour rate limits, citing new comp…
Krutrim Launches Indic AI Model Supporting 22 Languages
Indian AI unicorn Krutrim unveils a multilingual language model covering all 22 scheduled Indian languages, challenging …
NTT Launches Tsuzumi 2.0 LLM for Japanese Enterprise
Japan's NTT unveils Tsuzumi 2.0, a large language model purpose-built for Japanese enterprise workloads with enhanced mu…
JD.com Unveils xLLM Speculative Inference Architecture
JD.com reveals its xLLM speculative inference system at AICon Shanghai, promising magnitude-level speedups for LLM infer…
Open-Source AI Models Now Rival GPT-4 on Key Benchmarks
Leading open-source LLMs from Meta, Mistral, and others now match or exceed proprietary systems across major benchmarks,…
Hugging Face Drops 405B Open-Source Model
Hugging Face releases a 405B parameter open-source model that matches GPT-4 and Claude on key benchmarks, reshaping the …
Meta Llama 4 Maverick Beats GPT-4o Benchmarks
Meta releases Llama 4 Maverick, an open-weight model that outperforms OpenAI's GPT-4o across key benchmarks, reshaping t…
Google DeepMind Unveils Gemini 2.5 Ultra
Google DeepMind launches Gemini 2.5 Ultra, its most powerful AI model featuring native multimodal reasoning across text,…
Claude 4 Opus Breaks Science Reasoning Record
Anthropic's Claude 4 Opus achieves unprecedented scores on graduate-level science benchmarks, surpassing GPT-4o and Gemi…
OpenAI Launches GPT-5 Enterprise With Agentic AI
OpenAI unveils GPT-5 Enterprise tier featuring advanced agentic workflow capabilities, targeting large-scale business au…