Gemma 4 26B A4B

gemma-4-26b-a4b · Google DeepMind

Multimodal MoE, ~3.8B active. Vision, reasoning and tools.

Served on NVIDIA B200 GPUs · Spain (EU)

Model source

Hugging Facegoogle/gemma-4-26B-A4B-it

Specifications

Parameters: 26B · 3.8B active MoE
Hardware: MIG B200
Context: 32,768 tokens
Quantization: NVFP4
Modalities: text
Capabilities: Reasoning · Tools

Pricing

input

0.12 €

output

0.30 €

cache read

0.012 €

€ / million tokens · VAT excluded

Example (curl)

Compatible with the OpenAI SDK. Change the base URL and API key.

curl https://api.gpuflow.ai/v1/chat/completions \
  -H "Authorization: Bearer $GPUFLOW_API_KEY" \
  -d '{ "model": "gemma-4-26b-a4b",
        "messages": [{ "role": "user", "content": "Hola" }] }'

Try it with 5 € free

Create your account, point your SDK at GPU Flow and call this model on NVIDIA B200 GPUs, in Spain.

start free see all models