Krutrim AI Launches 22-Language Model From India
Indian AI unicorn Krutrim releases a multilingual large language model supporting 22 official Indian languages, challeng…
📂 LLM News
Latest articles in LLM News
Claude Gets Persistent Memory Across Chats
Anthropic rolls out persistent memory for Claude, allowing the AI to remember user preferences and context across separa…
Meta Launches Llama 4 Scout With 10M Token Context
Meta releases Llama 4 Scout, an open-source model featuring a 10 million token context window and mixture-of-experts arc…
OpenAI Launches GPT-5 Turbo With Major Upgrades
OpenAI unveils GPT-5 Turbo featuring advanced reasoning, native multimodal capabilities, and significant API improvement…
Krutrim AI Launches Indic Language Models for 22 Languages
Indian startup Krutrim AI unveils large language models supporting all 22 officially recognized Indian languages, target…
Japan's Fugaku Supercomputer Trains Largest Domestic AI Model
RIKEN leverages the Fugaku supercomputer to train Japan's largest foundation AI model, advancing the nation's AI soverei…
Rakuten Launches 70B Open Source Japanese LLM
Japanese e-commerce giant Rakuten releases a 70 billion parameter open source language model optimized for Japanese lang…
Snowflake Arctic 2.0 Takes Aim at Big Tech LLMs
Snowflake launches Arctic 2.0, an enterprise-focused LLM designed to rival foundation models from OpenAI, Google, and Me…
IBM Granite 4.0 Targets Regulated Industry AI
IBM launches Granite 4.0 foundation models designed for heavily regulated industries like banking, healthcare, and gover…
Microsoft Phi-5 Beats GPT-4 on Edge Devices
Microsoft's new Phi-5 small language model delivers GPT-4-level performance while running locally on smartphones and lap…
Hugging Face Drops Open Reasoning Models to Challenge GPT-o1
Hugging Face releases open-weight reasoning models that match proprietary systems from OpenAI and Google on key benchmar…
Claude 4.5 Sonnet Tops SWE-Bench Full Benchmark
Anthropic's Claude 4.5 Sonnet sets a new state-of-the-art on SWE-Bench Full, outperforming GPT-4o and Gemini in real-wor…