Google Gemini & Vertex AI API Status, Latency & Hosted LLM Directory (2026)

What is the latency, uptime SLA, and model lineup of Google Gemini & Vertex AI?

Google Gemini & Vertex AI maintains an average response latency of 75ms TTFT and an operational uptime SLA of 99.99% across its global infrastructure endpoints (us-central1, us-east4, europe-west4, asia-east1). It currently hosts 10 tracked foundation models including Gemini 3.5 Flash-Lite, Gemma 4 31B IT, Gemini 2.0 Flash, Llama 4 Maverick.

Verified daily via automated API latency tests and official documentation.

Zero-Shot RAG Verified

Average TTFT Latency 75ms P90 Benchmark Window

Historical Uptime SLA 99.99% 30-Day Rolling Telemetry

Regional Deployment

us-central1 us-east4 europe-west4 asia-east1

Multi-AZ Availability

Models Hosted on Google Gemini & Vertex AI (10)

Gemini 3.5 Flash-Lite

Google · Ultra-Compact Distilled MoE

81.5 IVR

Google's lowest-cost 1M-token model engineered for massive-scale classification, entity extraction, and sub-80ms real-time edge inference.

Input / Out $0.075 / $0.30

Host TTFT 78ms

Throughput 265 t/s

Gemma 4 31B IT

Google · Dense Open-Weight Transformer

80.3 IVR

Google's flagship Gemma 4 31B dense open-weight model delivering frontier-class coding and reasoning on a single enterprise GPU.

Input / Out $0.12 / $0.36

Host TTFT 115ms

Throughput 155 t/s

Gemini 2.0 Flash

Google DeepMind · Dense Multimodal Transformer

77.7 IVR

Previous-generation high-throughput 1M context multimodal workhorse (superseded by Gemini 3.8 Flash and Gemini 3.5 Flash-Lite).

Input / Out $0.10 / $0.40

Host TTFT 280ms

Throughput 185 t/s

Llama 4 Maverick

Meta · Multimodal Sparse MoE (17B Active / 400B Total)

77 IVR

Meta's flagship 128-expert open-weight multimodal MoE model delivering 1M context and rapid LPU/CS-3 inference throughput.

Input / Out $0.20 / $0.60

Host TTFT 150ms

Throughput 183.8 t/s

Gemini 3.8 Flash

Google · Hybrid Reasoning Multimodal MoE

74 IVR

High-velocity 1M-context multimodal reasoning model balancing sub-110ms TTFT latency with deep agentic tool execution and configurable thinking levels.

Input / Out $0.25 / $1.50

Host TTFT 105ms

Throughput 210 t/s

Gemini 2.5 Flash

Google · Hybrid Reasoning Omni Transformer

66.6 IVR

Hybrid reasoning model delivering strong price-performance across 1M context with configurable thinking budgets.

Input / Out $0.30 / $2.50

Host TTFT 110ms

Throughput 180 t/s

Gemini 4 Argon

Google · Next-Gen Sparse MoE with Adaptive High Thinking

62.6 IVR

Google DeepMind's flagship Gemini 4 Argon model featuring configurable high-depth reasoning, 1M token native multimodal context, and industry-leading agentic tool synthesis.

Input / Out $2.00 / $10.00

Host TTFT 850ms

Throughput 115 t/s

Gemini 3.1 Pro

Google · Deep Reasoning Multimodal MoE

60 IVR

Enterprise-grade 1M-token reasoning model optimized for complex repository analysis, multi-hour video comprehension, and custom tool synthesis.

Input / Out $1.25 / $10.00

Host TTFT 420ms

Throughput 98 t/s

Gemini 2.5 Pro

Google · Native Multimodal MoE with Thinking

56.8 IVR

Million-token multimodal model capable of analyzing hours of video, audio streams, and massive codebase repositories.

Input / Out $1.25 / $10.00

Host TTFT 700ms

Throughput 80 t/s

Claude 4.5 Sonnet

Anthropic · Extended Thinking Hybrid Transformer

56.5 IVR

Anthropic's premier software engineering and autonomous computer-use model featuring interleaved thinking and 90% prompt cache read discounts.

Input / Out $3.00 / $15.00

Host TTFT 405ms

Throughput 80.6 t/s

Frequently Asked Questions About Google Gemini & Vertex AI

What is the average latency and uptime SLA of Google Gemini & Vertex AI?

Google Gemini & Vertex AI delivers an average Time to First Token (TTFT) latency of 75ms and a verified 30-day rolling uptime SLA of 99.99%.

Which AI models are available on Google Gemini & Vertex AI?

Google Gemini & Vertex AI hosts 10 tracked LLMs on InferenceRate: Gemini 3.5 Flash-Lite, Gemma 4 31B IT, Gemini 2.0 Flash, Llama 4 Maverick, Gemini 3.8 Flash, Gemini 2.5 Flash, Gemini 4 Argon, Gemini 3.1 Pro, Gemini 2.5 Pro, Claude 4.5 Sonnet.

Which cloud regions does Google Gemini & Vertex AI support?

Google Gemini & Vertex AI supports API inference deployments across us-central1, us-east4, europe-west4, asia-east1.