Model by Google

Gemma 4 26B A4B

Gemma 4 26B A4B is Google DeepMind's mid-sized open model, and it suits people who want image and video understanding with low active compute under Apache 2.0.

Checked 2026-11-05

Gemma 4 26B A4B is Google DeepMind's mid-sized open model, and it suits people who want image and video understanding with low active compute under Apache 2.0. The Hugging Face card lists 25.2B total parameters with 3.8B active, 128 experts, and a context of up to 256K tokens. It takes text and images as input and treats video as frames of up to 60 seconds; there is no audio. Output is text. OpenRouter lists it from 3 April 2026 with a 262,144-token window at $0.025 per million input tokens and $0.075 per million output tokens on its cheapest route, checked on 5 November 2026. The card says the pre-training data stops in January 2025 and notes the usual limits on factual accuracy and bias. The card carries a technical report dated July, after the OpenRouter listing, so we treat the April date as a listing date only. Figures come from the Hugging Face card and the OpenRouter model list. We have not run it ourselves. We would use it as a cheap vision-capable model for local or low-cost work.

Mentioned here

Questions

What is the context window of Gemma 4 26B A4B?

262,144 tokens, with up to 235,929 tokens of output.

How much does Gemma 4 26B A4B cost?

$0.025 per million input tokens and $0.075 per million output tokens, checked 5 November 2026.

Can I run Gemma 4 26B A4B on my own hardware?

Yes, the weights are published under the Apache 2.0 licence. They are on huggingface.co.

Who makes Gemma 4 26B A4B?

Google makes it, and released it on 3 April 2026.

Reviews

Loading reviews…

Write a review

Your rating

Up to 1000 characters. Stars alone are enough.

Similar

  • Gemini 3.1 Flash Lite

    Gemini 3.1 Flash Lite is Google's cheap, high-volume multimodal model, suited to people who need low-latency work such as classification, extraction and light agent steps at the lowest cost.

    • Closed

    Google, 1M context, $0.25 in / 1M

  • Gemini 3.5 Flash

    Gemini 3.5 Flash is Google's fast multimodal model for people who wanted strong coding and agent work at a Flash-tier price, and it is now superseded.

    • Closed

    Google, 1M context, $1.50 in / 1M

  • Gemini 2.5 Flash

    Gemini 2.5 Flash is Google's June 2025 workhorse model with a 1M token context, and it is deprecated, so treat it as a model to migrate off.

    • Closed

    Google, 1M context, $0.30 in / 1M

  • Gemini 2.5 Pro

    Gemini 2.5 Pro is Google's June 2025 reasoning model with a context of 1,048,576 tokens, and it is now deprecated, so it only matters for code that still calls it.

    • Closed

    Google, 1M context, $1.25 in / 1M

  • Gemini 3.1 Pro Preview

    Gemini 3.1 Pro Preview is Google's older Pro-tier reasoning model, and it suits developers who need long-context multimodal work and accept preview terms.

    • Closed

    Google, 1M context, $2 in / 1M

  • Gemini 3.6 Flash

    Gemini 3.6 Flash is Google's fast, low-cost multimodal model for coding and agent work, and it suits teams with high-volume jobs that can live with a previous-generation model.

    • Closed

    Google, 1M context, $0.75 in / 1M