Bobbin
← Models

Gemma 4 26B A4B

26B of knowledge, a fraction of the work per word

googleApache-2.0FreeReleased March 2026 · 6 months old
Install with Bobbin

Don't have the app? Download it here.

A mixture-of-experts model: it holds 26B parameters but only uses a small slice for each word, so it answers much faster than a dense model its size.

It still needs the memory for all of it — the speed is free, the memory is not.

Available versions

QuantizationSizeVRAMRAMContext
Q4_K_M16.9 GB16.5 GBest.25.4 GB8kInstall

Memory needed grows with the length of a conversation, so these figures are for the context shown. Anything marked est. is worked out from the file size rather than measured on real hardware — the app says so there too. The desktop app checks all of it against your actual machine before installing.