Free models · data of 2026-11-06
Gemma 4 26B A4B: free API
Gemma 4 26B A4B is free on OpenRouter as google/gemma-4-26b-a4b-it:free. Today it ranks #1 of 12 free chat models in our open ranking, with a score of 56/100.
- Score basis
- Published benchmarks and LM Arena
- Served by
- Google AI Studio
- Context
- 262K tokens
- Max output
- 33K tokens
- Input
- image, text, video
- Tool calling
- Yes
- JSON output
- No
- Free limit
- 20 RPM · 50 RPD
Does it train on your prompts?
The free variant is served by Google AI Studio, which does not train on prompts and keeps them for up to 55 days.
From OpenRouter's published provider policies, checked 2026-10-22.
How good is it?
- Coding index
- 39.3
- Agentic index
- 2.9
- LM Arena (overall)
- 1434
- LM Arena (coding)
- 1446
- LM Arena (instructions)
- 1420
- LM Arena (Spanish)
- 1417
- Arena votes
- 6,090
- Parameters
- 25.8B
The indexes are Artificial Analysis benchmarks as published by OpenRouter. Arena ratings are human preference votes from LM Arena. How they combine into the score is in the ranking's scoring code.
How to call it
Any OpenAI-compatible client works. Create a key at openrouter.ai and use the model id above; the free limit is shared by all free models on your account.
from openai import OpenAI
client = OpenAI(base_url="https://openrouter.ai/api/v1", api_key="sk-or-...")
reply = client.chat.completions.create(
model="google/gemma-4-26b-a4b-it:free",
messages=[{"role": "user", "content": "Hello"}],
extra_body={"provider": {"data_collection": "deny"}}, # never route to a provider that may train
)
print(reply.choices[0].message.content)Hit a limit? free-models-per-day, Provider returned error and other errors, explained. Limits and fine print: OpenRouter free models.
Questions
- Is Gemma 4 26B A4B free to use?
- Yes, as google/gemma-4-26b-a4b-it:free on OpenRouter. Free models share one allowance per account: 20 RPM · 50 RPD, or 1,000 a day once the account has bought 10 credits.
- Does Gemma 4 26B A4B train on my prompts?
- The free variant is served by Google AI Studio, which does not train on prompts and keeps them for up to 55 days.
- What is Gemma 4 26B A4B's context window?
- 262K tokens of context, with up to 33K tokens of output.
- How good is Gemma 4 26B A4B compared with other free models?
- It ranks #1 of 12 free chat models today with a score of 56/100 (basis: Published benchmarks and LM Arena).
Ranked next to it
Sources
- Gemma 4 26B A4B on OpenRouter
- LM Arena leaderboard dataset
- The open ranking on GitHub (data/models.json)
Data of 2026-11-06, refreshed daily. Free models come and go: when this one leaves the free catalogue, this page goes with it.
Don't want to pick one model and handle its errors? ZeroLimitAI is one OpenAI-compatible endpoint that sends each request to the best free model available, and can run on your own Groq or OpenRouter key.