Blog

Google DeepMind releases Gemini 2.5 Pro, a multimodal AI model achieving #1 on LMArena with 1M token context and native reasoning.

DeepSeek releases the V3-0324 update, a massive MoE upgrade that crushes benchmarks in coding, math, and reasoning while remaining fully open-source.

NVIDIA has released Nemotron Ultra, an open-source 253B MoE model designed for complex reasoning and enterprise workloads.

Mistral AI releases Small 3.1, integrating vision capabilities into a 24B parameter model with 128K context window and permissive Apache 2.0 licensing.

Cohere releases Command A, a 111B parameter open-source model optimized for enterprise RAG and agentic workflows, running efficiently on just 2 GPUs.

Google DeepMind launches Gemma 3, a new family of multimodal models featuring 128K context and single-GPU efficiency under Apache 2.0.

Shanghai AI Lab unveils InternLM 3, an 8B parameter model surpassing Llama 3.1 on reasoning tasks with 128K context and Apache 2.0 licensing.

Alibaba Cloud releases QwQ-32B, a dedicated reasoning model with strong mathematical capabilities under an Apache 2.0 license. Explore benchmarks and pricing.

OpenAI unveils GPT-4.5, setting new standards for emotional intelligence and reduced hallucinations in enterprise AI applications.