NewsGoogle2 min read
Gemini 4 Argon Is Now Available: API, App and Prices
Gemini 4 Argon reaches the API worldwide and the Gemini app on Google AI Pro and Ultra. Launch price until 30 November and what else is new.
Key points
- Since 13 October any developer can use Gemini 4 Argon on the paid API, worldwide.
- In the Gemini app and AI Mode in Search, for Google AI Pro and Ultra subscribers.
- Launch price until 30 November: $2 per million input tokens and $10 output; then $4 and $20.
- Google now describes it as a 1 million token limit shared between input and output.
⚠ Pending verification. We have not yet confirmed these details with an official source. Check them before relying on them:
- Exact availability in the European Union: Google says “globally” but names no countries or languages.
Two weeks ago Google announced Gemini 4 Argon, and only a few cyber defenders could use it (see the launch story). On 13 October that changed: the model is rolling out worldwide on the API and to paying subscribers in the Gemini app.
Where you can use it
| Who | Where |
|---|---|
| Developers | Paid API in Google AI Studio and the Gemini platform; also Android Studio |
| Google AI Pro and Ultra subscribers | Gemini app and AI Mode in Search |
| Google AI Ultra subscribers | Google Antigravity, its environment for coding with agents |
| Businesses | Gemini for business, the API and Antigravity |
It is not in the free Gemini app.
What it costs on the API
| Until 30 November 2026 | From 1 December | |
|---|---|---|
| Input (per million tokens) | $2 | $4 |
| Output (per million tokens) | $10 | $20 |
Repeated input tokens that are cached get a 95% discount: if you send the same instructions or the same document again and again, you pay far less (see prompt caching). At the launch price it costs the same as GPT-6.1 Sol; once it rises, it moves closer to each company’s most expensive models.
The 1 million token limit, clarified
In the September announcement Google spoke of a 1 million token output limit. On the 13 October page it describes a 1 million token limit shared between input and output on the API, plus a “continuation protocol” for very long answers that developers have to set up. In practice you can give it a lot of text or ask for a lot of text, but the two together cannot go past a million. If that sounds odd, the glossary explains the difference between the context window and the output limit.
Other news from the same day
- Skills and scheduled tasks in the Gemini app: saved instructions you can reuse, and jobs that run on their own at the time you choose.
- Robotics: Google DeepMind published a version of Argon for robots. Google says it scores 87% on ERQA, a test of reasoning about the physical world.
- Safety: Google says that on Gray Swan’s indirect prompt injection benchmark, attacks succeeded only 0.7% of the time after 15 attempts.
- Science: 10,000 scientists can apply for 6 free months of Antigravity’s highest usage tier, subject to capacity and selection.
All the performance figures come from Google; there is no independent evaluation yet.
What it means for you
- If you pay for Google AI Pro, you can already pick Argon in the Gemini app and compare it with what you use now.
- If you build on the API, the launch price is a good window to try it before it doubles in December. To choose between models, see the comparison.
- If you are in Europe, check in AI Studio that it is enabled on your account: Google gives no per-country details.
