Gemma 4 26B A4B
Gemma 4 26B A4B is a large language model made by Google, added to the catalog on 3 Apr 2026. It has a context window of 262,144 tokens and costs $0.0675 per 1M input tokens and $0.225 per 1M output tokens. It accepts text, images and video and produces text. Typical uses: summarising long documents, answering questions about supplied material and interpreting image and video content.
Price checked ยท source: OpenRouter
What Gemma 4 26B A4B is
Google Gemma 4 26B A4B is an instruction-tuned Mixture-of-Experts (MoE) model from Google DeepMind. It accepts text, images, and video as input and produces text. The catalog lists a context length of 262,144 tokens and a maximum output length of 235,929 tokens. It identifies 25.2B total parameters and says 3.8B activate per token during inference. Listed capabilities include reasoning, tools, structured outputs, and response-format controls.
These specifications make the model relevant to tasks where a prompt combines lengthy text with visual input, or where the response needs to follow a requested structure. It may suit summarising documents, answering questions based on supplied text or media, and producing text in a structured format. Its video input can be included in workflows that ask for text about video content, while its image input can support prompts involving visual material. The catalog does not specify particular image or video processing features, so the scope of those tasks should not be assumed beyond the listed input modalities.
Tool support makes it suitable for applications that connect model responses with available tools, while reasoning support is relevant to reasoning-focused prompts. The 262,144-token context provides room for extensive input, although prompts and outputs must fit their respective limits. The catalog does not identify particular supported tools, output schemas, or response formats, so those depend on the implementation using the model.
Key features
- Accepts text, image, and video inputs
- Produces text
- Has a 262,144-token context window
- Supports tool use and reasoning
- Supports structured outputs and response-format controls
Typical uses
- Summarising long documents
- Answering questions about supplied material
- Interpreting image and video content
- Generating structured text responses
- Tool-assisted workflows
Gemma 4 26B A4B pricing
US dollars, list price through OpenRouter.
| Item | Standard | Free |
|---|---|---|
| Input, per 1M tokens | $0.0675 | $0 |
| Output, per 1M tokens | $0.225 | $0 |
| Cached input read, per 1M tokens | $0.0375 | n/a |
| Variant | Model id |
|---|---|
| Free | google/gemma-4-26b-a4b-it:free |
A variant is the same model under another id, with prices or limits of its own. Its prices are the columns of the list above.
The price last changed on 20 Nov 2026. One price change has been recorded since Gemma 4 26B A4B was added.
| Date | Input /1M | Output /1M | Change |
|---|---|---|---|
| $0.0675 | $0.225 | Price lowered | |
| $0.09 | $0.30 | Added at this price |
| What a task costs | Cost |
|---|---|
| Answer a short question500 in / 300 out | $0.0001 |
| Summarise a 50-page PDF25,000 in / 800 out | $0.0019 |
| Review a 2,000-line source file30,000 in / 2,000 out | $0.0025 |
| Read a 300-page contract set150,000 in / 1,500 out | $0.010 |
Gemma 4 26B A4B specifications
| Model id | google/gemma-4-26b-a4b-it |
|---|---|
| Version | gemma-4-26b-a4b-it-20260403 |
| Context window | 262,144 tokensAbout 196,608 English words |
| Max output | 235,929 tokens |
| Knowledge cutoff | Not published |
| Added to catalog | 3 Apr 2026 |
| Input types | image, text, video |
|---|---|
| Output types | text |
| Tool calling | Yes |
| Structured outputs | Yes |
| Reasoning effort control | No |
| Content moderation | Not applied by provider |
Sample outputs
The same prompts are given to every model, and the unedited output is shown with the date it was produced. Upend does not publish leaderboard scores it did not run.
No output has been recorded for Gemma 4 26B A4B yet.
Compare Gemma 4 26B A4B
Side by side with other models: prices, context window, what each accepts and supports, and the same four tasks priced on both.