Gemini 4 Argon in Langdock

Gemini 4 Argon is the model Google describes as its efficient Gemini 4 reasoning model. In Langdock it is available with hosting in the EU. You can use it in chat, in agents and workflows, and via the API, where it costs €2 per 1 million input tokens and €10.02 per 1 million output tokens.

Written by LangdockUpdated

What is Gemini 4 Argon good at?

Google names these typical uses for Gemini 4 Argon[1]:

  • Complex coding
  • Agentic tasks
  • Maths and science tasks
  • Work where response time and token spend matter

This is how Google describes the model, not a test result of ours. Which tools the model can use in Langdock depends on the tools and integrations enabled in your workspace.

What are the key facts for Gemini 4 Argon in Langdock?

Gemini 4 Argon: region, price and limits
FactValue
ProviderGoogle[4]
Name in the model selectorGemini 4 Argon[4]
Hosting region in LangdockEU[4]
Price per 1M input tokens€2[4]
Price per 1M output tokens€10.02[4]
Context window documented by Google1,048,576 tokens (input and output together)[1]
Maximum output documented by Google65,536 tokens by default[1]
Input types documented by GoogleText, images, video, audio and PDF[1]

The limits Google documents apply to its own API; Langdock can set its own. The context window indicator next to the model selector shows the window size in use.

Token prices apply to the API, to model use in workflows and to extra usage, excluding VAT. They are converted from US dollars at the current exchange rate, so they can change slightly from day to day. In chat and agents, Core and Work seats with included models have no usage-based costs; pricing lists the plans.

When should you choose Gemini 4 Argon, and when another model?

Google describes Gemini 4 Argon as delivering Gemini 4 intelligence with faster responses and lower token spend per turn than Gemini 4 Carbon, across complex coding, agentic, maths and science tasks. Gemini 4 Carbon is the model Google calls its most capable reasoning model, so it is the alternative when Argon’s results are not enough.[1][2][3]

The table compares what Langdock lists for Gemini 4 Argon and for the alternatives. The model list has every model.

Gemini 4 Argon and alternatives in Langdock
ModelProviderHosting regionInput / output per 1M tokens
Gemini 4 ArgonGoogleEU€2 / €10.02
Gemini 4 CarbonGoogleEU€2 / €10.02
Claude Sonnet 5.5AnthropicEU€2 / €10.02
GPT-6.1 SolOpenAIEU€2 / €10.02

How do you use Gemini 4 Argon in Langdock?

  • Chat: choose Gemini 4 Argon in the model selector of the chat input. You can switch models in the middle of a conversation and set a personal default model.
  • Agents: choose the model in the agent editor.
  • API: the Completion API has an endpoint that follows Google’s own API format, with the region in the path. Its models endpoint lists the models available to your workspace in that region. The Agents API lists the models you can use for agents.
  • Admins: workspace admins enable or disable each model in the workspace settings and choose the default model for new chats and agents. If you do not see the model, ask your admin.

Details are in the model guide of the documentation.

Where is Gemini 4 Argon hosted, and what happens to your data?

The Langdock model list shows the hosting region EU for Gemini 4 Argon[4]. In a new workspace only models hosted in the EU are enabled by default; admins decide which other models are switched on[9].

For all models, Langdock states that customer data is never used to train or improve AI models[10], and it publishes a data processing agreement (DPA)[11]. A hosting region alone does not make a use case GDPR-compliant: the guide to GDPR-compliant AI explains what Langdock provides and what remains your responsibility.

Frequently asked questions

Yes. The Langdock model list shows the hosting region EU for Gemini 4 Argon.

Via the API, in workflows and as extra usage, Gemini 4 Argon costs €2 per 1 million input tokens and €10.02 per 1 million output tokens, excluding VAT. In chat and agents, Core and Work seats with included models have no usage-based costs.

Google documents a context window of 1,048,576 tokens (input and output together) for its own API. Langdock can set its own limit; the context window indicator next to the model selector shows the size in use.

No. Langdock states that customer data is never used to train or improve AI models.

Sources

  1. Google: Gemini 4 Argon model page (checked on )
  2. Google: Gemini 4 Carbon model page (checked on )
  3. Google: Gemini models overview (checked on )
  4. Langdock: Models with hosting region
  5. Langdock documentation: Model guide
  6. Langdock documentation: Agent configuration
  7. Langdock documentation: Completion API
  8. Langdock documentation: Models for the Agents API
  9. Langdock documentation: Workspace settings, models
  10. Langdock: Security
  11. Langdock: Data processing agreement (DPA)

Gemini 4 Argon is a product of Google. Product names are trademarks of their owners. Statements about the model were taken from Google’s documentation on the date shown in the sources.