Gemma 4 26B A4B

Price

Made by

Google

Input

$0.12 per million tokens

Output

$0.36 per million tokens

Context

256K tokens

Can

Tools, Reasoning, JSON, Reads images

Use it from your code

GridPup's API speaks OpenAI's format: use any OpenAI SDK with the base URL https://api.gridpup.com/v1/ai and a GridPup API key with Also use AI models. The answer tells you what it cost. More in the API docs.

from openai import OpenAI

client = OpenAI(base_url="https://api.gridpup.com/v1/ai", api_key="gp_live_…")

r = client.chat.completions.create(
    model="@cf/google/gemma-4-26b-a4b-it",
    messages=[{"role": "user", "content": "Explain a KV cache in two sentences."}],
)
print(r.choices[0].message.content)
print("cost:", r.usage.cost_usd, "USD")

Questions

How much does Gemma 4 26B A4B cost?

$0.12 per million input tokens and $0.36 per million output tokens, charged per request from your prepaid balance. You need a small balance to start ($1 for most models); there is no monthly fee.

Who runs it?

Google made Gemma 4 26B A4B. GridPup serves it through Cloudflare's AI Gateway, so you use it with one GridPup balance instead of an account with Google.

Does it work with the OpenAI SDK?

Yes: set base_url to https://api.gridpup.com/v1/ai and the model to @cf/google/gemma-4-26b-a4b-it.

Similar chat and reasoning models

• Llama 3.2 1BMeta$0.033 per million input tokens
• Llama 3.1 8BMeta$0.18 per million input tokens
• gpt-oss 20B on GroqOpenAI$0.091 per million input tokens
• gpt-oss 20BOpenAI$0.24 per million input tokens
• Llama 3.2 3BMeta$0.061 per million input tokens
• Qwen3 30B A3BQwen$0.061 per million input tokens

All chat and reasoning models and prices