Model Releases
Meta Unveils Llama 3: The New Open-Source Standard
Meta releases Llama 3, a 70B parameter model trained on 15T tokens, setting a new benchmark for open-source AI.
5 min read
Latest news, tutorials, and insights about AI
Microsoft releases Phi-3, a 14B parameter open-source model rivaling larger competitors with phone-capable efficiency.
Meta releases Llama 3, a 70B parameter model trained on 15T tokens, setting a new benchmark for open-source AI.
Mistral AI releases Mixtral 8x22B, a 176B parameter open-source Mixture of Experts model with superior multilingual and coding capabilities.
Cohere releases Command R+, a 104B parameter open-source model optimized for enterprise RAG applications with 128K context window.
AI21 Labs releases Jamba 52B, the world's first production Mamba-Transformer hybrid model with 256K context window and novel SSM architecture.
Databricks releases DBRX, a 132B parameter Mixture of Experts model with 36B active parameters, outperforming Llama 2 70B and Mixtral under Apache 2.0 license.
xAI releases Grok-1 as its first open-source model under Apache 2.0 license, featuring 314B parameters with Mixture of Experts architecture.
Anthropic's Claude 3 family revolutionizes AI with unprecedented reasoning, coding, and multimodal capabilities that match or exceed GPT-4 benchmarks.
Anthropic's Claude Opus 3 represents a historic leap in AI reasoning, featuring 200K context window and advanced multimodal capabilities.