Records found: 7
'AI2 releases Olmo 3, an open 7B and 32B LLM family built on the Dolma 3 data suite with a 65,536 token context and fully transparent training pipeline. Variants include reasoning, instruction-following, and RL-ready models for reproducible research.'
K2 Think is a 32B open-weight reasoning system from MBZUAI that matches or outperforms much larger open models on math by combining long CoT fine-tuning, RL with verifiable rewards, planning, and wafer-scale speculative decoding.
'Alibaba introduces two 4B-parameter models with native 256K token context windows: an Instruct model for concise multilingual responses and a Thinking model for explicit chain-of-thought reasoning, both optimized for consumer deployments.'
OpenAI’s new GPT-5 model offers faster reasoning, better user experience, and fewer hallucinations, but represents a refinement rather than a breakthrough on the path to AGI.
DeepSeek released R1-0528, an open-source reasoning AI model with improved math and code performance that runs efficiently on a single GPU, challenging top industry models.
NVIDIA introduces Llama Nemotron Nano 4B, a compact open-source AI model optimized for edge deployment that outperforms larger models in scientific reasoning and programming tasks.
PrimeIntellect has launched INTELLECT-2, a 32-billion parameter reasoning model trained using decentralized asynchronous reinforcement learning, achieving superior performance and open-sourcing the full codebase.