# DiffusionGemma 26B A4B

> DiffusionGemma 26B A4B IT is an open-weights multimodal model from Google DeepMind that generates text via discrete diffusion. Built on the Gemma 4 26B A4B MoE architecture (25.2B total / 3.8B active), it emits tokens in parallel 256-token blocks for high-throughput generation, with a 256K context window, configurable thinking mode, native function calling, and 35+ language support.

**ID**: `google/diffusiongemma-26b-a4b-it`  
**Creator**: Google  
**Category**: text  
**Context**: 262K tokens  
**Status**: Disabled (since 2026-08-19) — no longer routable or listed; kept for history/usage reference.  
**Reason**: Chat completions return empty content for this discrete-diffusion model under normal chat token budgets. It is unlisted rather than advertised as a live chat model.  
**Input**: $0  
**Output**: $0  
**Released**: 2026-06-10  
**Web page**: https://anyrouter.dev/model/google/diffusiongemma-26b-a4b-it

**Input modalities**: text, image, video  
**Output modalities**: text  

**Tokenizer**: Gemma

**Capabilities**: chat, reasoning, streaming, function-calling, vision, coding

**Supported parameters**: max_tokens, temperature, top_p, stop

## Usage

**Endpoint**: `POST https://anyrouter.dev/api/v1/chat/completions`

```bash
curl https://anyrouter.dev/api/v1/chat/completions \
  -H "Authorization: Bearer $ANYROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "google/diffusiongemma-26b-a4b-it",
  "messages": [
    {
      "role": "user",
      "content": "Say hi in 3 words."
    }
  ]
}'
```

## Providers

| Provider | Input | Output |
| --- | --- | --- |
| nvidia | $0 | $0 |
| nvidia-byok | $0 | $0 |

## Benchmarks

- **GPQA Diamond**: 0.7
- **HLE**: 0.1
- **SciCode**: 0.3