Blog

Google DeepMind releases Gemini 2.5 Flash, combining top-tier reasoning with unmatched speed and cost efficiency for enterprise workloads.

Mistral AI unveils Mistral Medium 3, a front-tier Apache 2.0 model competing with GPT-4o, featuring strong multilingual support and edge deployment capabilities.

DeepSeek releases DeepSeek-Prover-V2, a second-generation open-source theorem prover for Lean 4 that sets new state-of-the-art benchmarks in formal mathematics.

Alibaba Cloud launches Qwen 3, a 235B parameter MoE model with 119 language support and Apache 2.0 licensing. Discover the specs, benchmarks, and deployment guide.

Zhipu AI unveils GLM-4.1V, a 32B parameter multimodal model offering competitive vision capabilities and open-source reasoning.

OpenAI releases o4-mini, a specialized reasoning model optimized for cost-performance in coding and STEM tasks on April 16, 2025.

OpenAI unveils o3, a successor to o1 designed for complex chain-of-thought reasoning, offering enhanced capabilities for professional workflows.

OpenAI releases the GPT-4.1 Series on April 14, 2025, featuring a massive 1M token context window and specialized variants for developers.

Meta AI releases Llama 4, featuring 400B+ MoE architecture and native multimodal fusion. A historic shift for open-weight AI.