Google: Gemini 4 Hydrogen is the high-performance workhorse model in the Google Gemini 4 series. It is optimized for high-throughput proxy workflows, encoding, and multi-model routing. It is lower cost and has lower latency than Gemini 4 Carbon. It supports text, image, audio, video and file input, supports long text generation and can continue to write up to 1 million output tokens, and provides four inference depth settings of low, medium, high and ultra-high to achieve controllable inference depth.
Success rate of requests sent by the Amux AI gateway to each provider. Client errors are excluded.
Tok/s
24 hours · 11/16–11/21
Average output speed of streaming requests, in tokens per second (TPS).
TTFT
24 hours · 11/16–11/21
Average time to first token (TTFT) of streaming requests, measured from request dispatch to the first token received.
Activity
24 hours · 11/16–11/21
Relative request activity over time, broken down by serving provider. Indexed to the peak within the selected range (= 100); reflects trends and provider share, not absolute volume.
Related models
More models from Google
Google: Gemini 4 Carbon1.05M contextfrom $0.4 / M tokens2 providersGoogle: Gemini 4 Carbon is a cutting-edge artificial intelligence model from Google DeepMind built on Gemini 4 Argon, designed to handle complex tasks that require continuous inference and multi-step execution, including debugging large code bases, cybersecurity defense, scientific research, and professional data analysis. It supports text, image, audio and video input, and has a comprehensive input and output context window of approximately 1 million tokens.
Google: Gemini 4 Argon1.05M contextfrom $0.4 / M tokens2 providersGoogle: Gemini 4 Argon is a cutting-edge model released by Google DeepMind on September 30, 2026. It is the first model in the Gemini 4 series to feature a new naming scheme. It is strong in programming, financial research, legal drafting, and autonomously fixing cybersecurity vulnerabilities. The model supports text, image, video, and file inputs, and has a context window of 1 million tokens. In benchmark tests, its DeepSWE v1.1 score was 77.9%, ahead of Claude Opus 5.5 and GPT-6 Astra, and ranked first in the Vals Index with 68.90%.
Google: Gemini 3.8 Flash1.05M contextfrom $0.15 / M tokens2 providersGemini 3.8 Flash is Google's most intelligent Flash model with significant gains from 3.7 Flash across software engineering, agentic tasks, and multi-step reasoning.