Google: Gemma 4 26B A4B (free)
🧠 AI Modelgoogle
Google's 26B MoE model with only 3.8B active parameters, free on OpenRouter.
Gemma 4 26B A4B IT is an instruction-tuned MoE model with 25.2B total parameters and 3.8B activated per token. It achieves competitive performance on benchmarks like MMLU, GSM8K, and HumanEval. Supports multimodal inputs (image, text, video) and outputs text. Features include frequency penalty, reasoning, response format, and seed. Context length of 262,144 tokens. Free tier available on OpenRouter.
💡Highlights
- ├─26B MoE, 3.8B active per token
- ├─262k context window
- └─Free on OpenRouter
🎯For
- ├─AI researchers
- ├─Developers
- └─Students