Gemma 4
2026-04-02 · Publisher : Google · Domain : Artificial Intelligence
Google DeepMind releases Gemma 4, its most capable family of open models, under the Apache 2.0 license, in four sizes (E2B, E4B, 26B MoE, 31B dense) designed for reasoning and agents.
What changes
- The 31B model ranks third among open models on the Arena AI text leaderboard, the 26B sixth, as of the date of the post.
- Native function calling, structured JSON output and native system instructions for agents.
- Vision (images and video) on all models; audio input on E2B and E4B.
- More than 140 languages, context from 128K (small models) to 256K (large ones).
- Apache 2.0 license.
### Specifications - Sizes: E2B (2.3B effective), E4B (4.5B), 26B A4B MoE (3.8B active), 31B dense. - Context: 128K tokens (E2B, E4B), 256K tokens (26B, 31B). - Availability: open weights and Gemini API (free access according to the pricing page). - Knowledge cutoff: January 2025 (pretraining data). ### Pricing Open weights under Apache 2.0. Free use through the Gemini API according to the pricing page. ### Benchmarks | Benchmark (model card, scores published by the vendor) | 31B | 26B A4B | E4B | E2B | |---|---|---|---|---| | MMLU Pro | 85.2% | 82.6% | 69.4% | 60.0% | | AIME 2026 | 89.2% | 88.3% | 42.5% | 37.5% | | LiveCodeBench v6 | 80.0% | 77.1% | 52.0% | 44.0% | | GPQA Diamond | 84.3% | 82.3% | 58.6% | 43.4% | | MMMU Pro (vision) | 76.9% | 73.8% | 52.6% | 44.2% | ### Key takeaways A credible open foundation for agents hosted in-house or on a workstation, with a permissive license that simplifies commercial use.
Links
Tags : LLM, Open Source, Google, Gemma, Agents