Term
Gemma 4
Gemma 4 is Google DeepMind's open model family from April 2026 — four sizes (E2B to 31B), 256k context, multimodal, Apache 2.0 licensed.
Gemma 4 — explained in more detail
Google DeepMind released Gemma 4 on April 2, 2026 as the most capable open model family in the Gemma line so far — built on the same research and technology as Gemini 3. Four sizes are available: Effective 2B (E2B), Effective 4B (E4B), 26B Mixture-of-Experts (with only 3.8B active parameters for low latency) and 31B Dense for maximum quality and as a fine-tuning foundation. Licensed under Apache 2.0 — free for commercial use.
Example / Practical use
Gemma 4 handles text and images (with audio supported on the small models) and covers more than 140 languages. The 256k context window, built-in reasoning mode and native function calling make it a basis for autonomous agents. On the Arena AI text leaderboard, the 31B model ranks third among open models and the 26B MoE sixth. On the AIME 2026 math benchmark, 31B scores 89.2%. Day-one support: Hugging Face, vLLM, llama.cpp, MLX, Ollama, NVIDIA NIM/NeMo, LM Studio, Unsloth, SGLang and many more — from cloud GPU to smartphone.
Entdecke mehr
Three Days with Claude Fable 5 — and Why Opus Stays My Daily Driver
Fable 5 fixed a bug in one shot that Opus failed at twice. I'm still not switching. My honest impressions from the trenches after three days.
GlossarMiniMax
MiniMax is an AI lab founded in Shanghai in 2021 whose open-weight M model family (LLMs) targets very long context windows and low running costs via sparse attention.
LexikonLLM Model Families at a Glance
Claude, GPT, Gemini, Llama & co. — who builds what, where each family shines, and how to pick the right model for your own use case.