Restricted defensive cybersecurity version of Argon for code analysis, vulnerability research and patching. Requires approval separate from Flash Cyber.
Argon Cyber is subject to Google Frontier Model terms. Prompts/responses may be logged inside the customer environment for up to 30 days, without training use or CMEK. Master Agreement customers are exempt from human review; online-terms customers may have flagged prompts reviewed. Google describes the former as ZDR-aligned, which does not mean no stored logs.Source for Data retention (opens in a new tab)
Shared ceiling minus input, using continuation mode.
Long output mode
GenerateContent long decode is supported only on v1beta1. SDK/Agent Studio can follow continuation_token automatically; custom clients may loop. Standard output is at most 65,536 tokens, or 32,768 for non-streaming direct HTTP.Source for Long output mode (opens in a new tab)
standard global rates, output includes thinking. Cyber variants do not receive the general-model introductory promotion. Long decode may bill replayed input; standard quotes cover one synchronous request. Cached input price: $0.20 / 1M tokens (base tier).
Argon Cyber is subject to Google Frontier Model terms. Prompts/responses may be logged inside the customer environment for up to 30 days, without training use or CMEK. Master Agreement customers are exempt from human review; online-terms customers may have flagged prompts reviewed. Google describes the former as ZDR-aligned, which does not mean no stored logs.
standard us rates, output includes thinking. Cyber variants do not receive the general-model introductory promotion. Long decode may bill replayed input; standard quotes cover one synchronous request. Cached input price: $0.22 / 1M tokens (base tier).
Argon Cyber is subject to Google Frontier Model terms. Prompts/responses may be logged inside the customer environment for up to 30 days, without training use or CMEK. Master Agreement customers are exempt from human review; online-terms customers may have flagged prompts reviewed. Google describes the former as ZDR-aligned, which does not mean no stored logs.
priority global rates, output includes thinking. Cyber variants do not receive the general-model introductory promotion. Long decode may bill replayed input; standard quotes cover one synchronous request. Cached input price: $0.36 / 1M tokens (base tier).
Argon Cyber is subject to Google Frontier Model terms. Prompts/responses may be logged inside the customer environment for up to 30 days, without training use or CMEK. Master Agreement customers are exempt from human review; online-terms customers may have flagged prompts reviewed. Google describes the former as ZDR-aligned, which does not mean no stored logs.
priority us rates, output includes thinking. Cyber variants do not receive the general-model introductory promotion. Long decode may bill replayed input; standard quotes cover one synchronous request. Cached input price: $0.396 / 1M tokens (base tier).
Argon Cyber is subject to Google Frontier Model terms. Prompts/responses may be logged inside the customer environment for up to 30 days, without training use or CMEK. Master Agreement customers are exempt from human review; online-terms customers may have flagged prompts reviewed. Google describes the former as ZDR-aligned, which does not mean no stored logs.
batch global rates, output includes thinking. Cyber variants do not receive the general-model introductory promotion. Long decode may bill replayed input; standard quotes cover one synchronous request. Cached input price: $0.10 / 1M tokens (base tier).
Argon Cyber is subject to Google Frontier Model terms. Prompts/responses may be logged inside the customer environment for up to 30 days, without training use or CMEK. Master Agreement customers are exempt from human review; online-terms customers may have flagged prompts reviewed. Google describes the former as ZDR-aligned, which does not mean no stored logs.
batch us rates, output includes thinking. Cyber variants do not receive the general-model introductory promotion. Long decode may bill replayed input; standard quotes cover one synchronous request. Cached input price: $0.11 / 1M tokens (base tier).
Argon Cyber is subject to Google Frontier Model terms. Prompts/responses may be logged inside the customer environment for up to 30 days, without training use or CMEK. Master Agreement customers are exempt from human review; online-terms customers may have flagged prompts reviewed. Google describes the former as ZDR-aligned, which does not mean no stored logs.
The generic cache guide instead gives 4,096 for Gemini 4. This host-specific discrepancy is unresolved; the model card is displayed. Developer API has a separate 4,096-token statement.
Common prompt prefixes may hit automatically. Disable durable caching alone or all implicit caching per project; explicit cache objects are unsupported.Source for How to enable it (opens in a new tab)
Published Argon token rates give a 95% cached-read discount, despite the generic guide describing 90%. Implicit storage has no separate charge.Source for Billing rules (opens in a new tab)
Limits & data boundaries
Durable KV data is stored at rest using Google encryption, not CMEK; project-isolated, never used for training, retained at most 24 hours. Regional endpoints keep cached data in the selected location; global requests/cache may be served from any region.Source for Limits & data boundaries (opens in a new tab)
standard global rates, output includes thinking. Cyber variants do not receive the general-model introductory promotion. Long decode may bill replayed input; standard quotes cover one synchronous request.
Observed 14 Oct 2026. Write rates and storage fees are different charges; use the billing rules above.
Caching reviewed 14 Oct 2026
Google Enterprise Agent Platform (standard, us) · standard
The generic cache guide instead gives 4,096 for Gemini 4. This host-specific discrepancy is unresolved; the model card is displayed. Developer API has a separate 4,096-token statement.
Common prompt prefixes may hit automatically. Disable durable caching alone or all implicit caching per project; explicit cache objects are unsupported.Source for How to enable it (opens in a new tab)
Published Argon token rates give a 95% cached-read discount, despite the generic guide describing 90%. Implicit storage has no separate charge.Source for Billing rules (opens in a new tab)
Limits & data boundaries
Durable KV data is stored at rest using Google encryption, not CMEK; project-isolated, never used for training, retained at most 24 hours. Regional endpoints keep cached data in the selected location; global requests/cache may be served from any region.Source for Limits & data boundaries (opens in a new tab)
standard us rates, output includes thinking. Cyber variants do not receive the general-model introductory promotion. Long decode may bill replayed input; standard quotes cover one synchronous request.
Observed 14 Oct 2026. Write rates and storage fees are different charges; use the billing rules above.
Caching reviewed 14 Oct 2026
Google Enterprise Agent Platform (priority, global) · priority
The generic cache guide instead gives 4,096 for Gemini 4. This host-specific discrepancy is unresolved; the model card is displayed. Developer API has a separate 4,096-token statement.
Common prompt prefixes may hit automatically. Disable durable caching alone or all implicit caching per project; explicit cache objects are unsupported.Source for How to enable it (opens in a new tab)
Published Argon token rates give a 95% cached-read discount, despite the generic guide describing 90%. Implicit storage has no separate charge.Source for Billing rules (opens in a new tab)
Limits & data boundaries
Durable KV data is stored at rest using Google encryption, not CMEK; project-isolated, never used for training, retained at most 24 hours. Regional endpoints keep cached data in the selected location; global requests/cache may be served from any region.Source for Limits & data boundaries (opens in a new tab)
priority global rates, output includes thinking. Cyber variants do not receive the general-model introductory promotion. Long decode may bill replayed input; standard quotes cover one synchronous request.
Observed 14 Oct 2026. Write rates and storage fees are different charges; use the billing rules above.
Caching reviewed 14 Oct 2026
Google Enterprise Agent Platform (priority, us) · priority
The generic cache guide instead gives 4,096 for Gemini 4. This host-specific discrepancy is unresolved; the model card is displayed. Developer API has a separate 4,096-token statement.
Common prompt prefixes may hit automatically. Disable durable caching alone or all implicit caching per project; explicit cache objects are unsupported.Source for How to enable it (opens in a new tab)
Published Argon token rates give a 95% cached-read discount, despite the generic guide describing 90%. Implicit storage has no separate charge.Source for Billing rules (opens in a new tab)
Limits & data boundaries
Durable KV data is stored at rest using Google encryption, not CMEK; project-isolated, never used for training, retained at most 24 hours. Regional endpoints keep cached data in the selected location; global requests/cache may be served from any region.Source for Limits & data boundaries (opens in a new tab)
priority us rates, output includes thinking. Cyber variants do not receive the general-model introductory promotion. Long decode may bill replayed input; standard quotes cover one synchronous request.
Observed 14 Oct 2026. Write rates and storage fees are different charges; use the billing rules above.
Caching reviewed 14 Oct 2026
Google Enterprise Agent Platform (batch, global) · batch
The generic cache guide instead gives 4,096 for Gemini 4. This host-specific discrepancy is unresolved; the model card is displayed. Developer API has a separate 4,096-token statement.
Common prompt prefixes may hit automatically. Disable durable caching alone or all implicit caching per project; explicit cache objects are unsupported.Source for How to enable it (opens in a new tab)
Published Argon token rates give a 95% cached-read discount, despite the generic guide describing 90%. Implicit storage has no separate charge.Source for Billing rules (opens in a new tab)
Limits & data boundaries
Durable KV data is stored at rest using Google encryption, not CMEK; project-isolated, never used for training, retained at most 24 hours. Regional endpoints keep cached data in the selected location; global requests/cache may be served from any region.Source for Limits & data boundaries (opens in a new tab)
batch global rates, output includes thinking. Cyber variants do not receive the general-model introductory promotion. Long decode may bill replayed input; standard quotes cover one synchronous request.
Observed 14 Oct 2026. Write rates and storage fees are different charges; use the billing rules above.
Caching reviewed 14 Oct 2026
Google Enterprise Agent Platform (batch, us) · batch
The generic cache guide instead gives 4,096 for Gemini 4. This host-specific discrepancy is unresolved; the model card is displayed. Developer API has a separate 4,096-token statement.
Common prompt prefixes may hit automatically. Disable durable caching alone or all implicit caching per project; explicit cache objects are unsupported.Source for How to enable it (opens in a new tab)
Published Argon token rates give a 95% cached-read discount, despite the generic guide describing 90%. Implicit storage has no separate charge.Source for Billing rules (opens in a new tab)
Limits & data boundaries
Durable KV data is stored at rest using Google encryption, not CMEK; project-isolated, never used for training, retained at most 24 hours. Regional endpoints keep cached data in the selected location; global requests/cache may be served from any region.Source for Limits & data boundaries (opens in a new tab)
batch us rates, output includes thinking. Cyber variants do not receive the general-model introductory promotion. Long decode may bill replayed input; standard quotes cover one synchronous request.
Observed 14 Oct 2026. Write rates and storage fees are different charges; use the billing rules above.
Caching reviewed 14 Oct 2026
What the measurements say
No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.
Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.
The record behind the model
Why it is included
Google names this as a distinct proprietary Argon version and API selector for defensive cybersecurity. No public weights are supplied. Separate architecture, parameter count or independent training are not established.
