Gemma 4 Didn't Just Get Smarter. It Became a Different Kind of Model. Here's What the Agentic Numbers Actually Mean. (opens in new tab)

This is a submission for the Gemma 4 Challenge: Write About Gemma 4 Every open-weight model release in 2026 comes with a benchmark table and a claim about efficiency. Most of them are incremental. Gemma 4 has one number that isn't: 6.6% to 86.4% on agentic tool use. That's not an improvement. That's a category change. The Number That Actually Matters When Google DeepMind dropped Gemma 4 on April 2, 2026, the coverage focused on the headline scores - AIME 2026, LiveCodeBench, Arena AI rankings...