Frontier builder · United Kingdom

Google DeepMind

Frontier IC25 · Edition 2026.3 · Evidence reviewed October 8, 2026 UTC

Gemini connects multimodal model research with a large consumer and cloud distribution platform.

Selected for a consequential, inspectable general-purpose model contribution. The displayed model is a named representative, not a claim of latest or best performance.

Research organization's London headquarters/origin; parent Google/Alphabet is US-based. Country is not API contracting entity or inference geography.

Selected model

Gemini 3.8 Flash · gemini-3.8-flash

Introductory first-party synchronous API rates through December 31, 2026; uncached text tokens, prompt ≤100K. Published January 1, 2027 rates are $1.50 input / $7.50 output per 1M tokens.

1,048,576 · Input only

Scope and limitations

• Listed standard rates are introductory and end December 31, 2026; published January 1, 2027 rates double to $1.50 input/$7.50 output.
• Output billing includes thinking tokens.
• The published 1,048,576 limit is an input-token limit. Flag this when plotting beside providers that report a combined context window.
• PDF is a provider-supported input document type, not a separate sensor modality.
• Country refers to the UK research organization; parent nationality is different.
• This selection does not describe Google's separate Gemma open-weight models.

Original sources

• Model input/output limits
• Standard paid pricing and scheduled increase
• API service terms
• Research organization London headquarters
• Tensor Processing Units (TPUs)
• Gemini Enterprise Agent Platform (formerly Vertex AI)
• Gemini: Your AI assistant from Google

Explore the full IC25 tool · Read the field guide in six languages