Model Families

Cohere's Aya Expanse is a multilingual model family optimized for 23 languages. Available in 8B and 32B sizes, these mod...

Deep Cogito's hybrid reasoning models that can dynamically switch between fast direct responses and deep chain-of-though...

Cohere's Command R is a family of models optimized for retrieval-augmented generation (RAG) and enterprise use cases. Co...

DeepSeek's R1 family of reasoning-focused open-weight models, trained with reinforcement learning to excel at complex mu...

DeepSeek's V3 series of mixture-of-experts models with 671B total parameters and 37B active per token. Among the most ca...

The third generation of TII's Falcon models, offering efficient 7B and 10B parameter variants. Designed for strong gener...

Google's Gemma 2 is a family of lightweight, open-weight models built from the same research and technology used to crea...

Google's Gemma 3 is a major upgrade over Gemma 2, featuring native multimodal support (text + image input) starting at 4...

Google's Gemma 4 is the most capable open model family from Google DeepMind, released April 2026 under Apache 2.0. It sp...

Zhipu AI's GLM family includes the GLM-5 flagship reasoning model — a 744B parameter MoE with 40B active parameters per ...

Meta's Llama 3 is one of the most capable and widely adopted open-weight model families. Spanning from compact 1B parame...

Meta's Llama 4 introduces mixture-of-experts architecture and native multimodal support to the Llama family. Scout (109B...

Mistral AI's open-weight model family, known for exceptional efficiency and strong performance relative to model size. I...

NVIDIA's Nemotron 3 family features novel hybrid architectures combining Mamba and Transformer blocks. Optimized for inf...

Microsoft's Phi family of small language models, designed to demonstrate that carefully curated training data can enable...

Alibaba's Qwen 2.5 is a comprehensive family of open-weight models spanning from 7B to 72B parameters, with specialized ...

Alibaba's Qwen 3 is the next generation of the Qwen family, featuring both dense models (0.6B to 32B) and mixture-of-exp...

Alibaba's Qwen 3.5 is a multimodal model family spanning 0.8B to 397B parameters, supporting 201 languages with 256K con...

BigCode's StarCoder is a family of code-specialized language models developed as part of an open scientific collaboratio...