Tasks and inputs
Gemini 4 Hydrogen is listed for Multimodal, Text generation, Coding & Code Generation. It accepts text, image, audio, video, pdf and produces text. Use the publisher’s input format and preprocessing for your chosen deployment.
SyncDev model directory · Profile updated Nov 17, 2026
Gemini 4 Hydrogen is Google's fast, lower-cost Gemini 4 model for everyday coding and agentic execution, with a 1M context, xhigh thinking and a free tier.
The specification score summarizes recorded features and limits. It is not a benchmark result.
Pricing conditions: Indicative paid-tier pricing per 1M tokens: $1.50 input / $7.50 output, flat across the 1M window; implicit caching $0.075; Batch/Flex $0.75 / $3.75; Priority $2.70 / $13.50. Grounding: 5,000 free requests/month, then $14 per 1,000. Free tier available. Confirm in your Google billing console.
Recorded modality support shows what this model can accept and produce. An undocumented modality is shown as unknown rather than assumed to be unsupported.
Accepts
Produces
Most Gemini traffic will never touch Carbon or Argon. It will hit whatever model sits behind an everyday coding assistant, a spreadsheet agent or a search answer, and from 12 November 2026 that model is Gemini 4 Hydrogen. Google calls it the daily driver for agentic workflows. At $1.50 per million input tokens and $7.50 per million output on the paid tier, with a free tier on the Gemini API, it is priced to be the default.
| Field | Value |
|---|---|
| API ID | gemini-4-hydrogen |
| Released | 12 November 2026 (GA) |
| Context window | 1,048,576 tokens |
| Input | Text, image, audio, video, PDF |
| Output | Text |
| Thinking levels | low, medium, high (default), xhigh |
| Max output | Up to 1M per Google's launch post; "64K / 1M" on the models page; 65,536 in third-party catalogues |
| Free tier | Yes (Gemini API) |
That last row needs care. Google's launch material says up to one million output tokens. Its model list gives "64K / 1M" for the whole Gemini 4 family. Third-party listings record 65,536. The safe reading is that 64K is the ordinary ceiling and longer output may require configuration, so test a long generation on your account before designing around it.
Earlier Flash models stopped at high. Hydrogen adds xhigh. Google reports an Artificial Analysis Intelligence Index score of 51 at high, edging Argon's 50, and 54 at xhigh. That is a publisher-reported comparison, but it hints at something useful: on some work, a cheap model thinking harder may match a pricier one thinking normally. Measure it on yours.
None of these were measured by us.
Hydrogen is widely described as 3.8 Flash's successor as Google's workhorse model. Google's own wording is softer: it frames Hydrogen as the upgrade for teams on 3.8 Flash, and no shutdown date for 3.8 Flash has been announced. So there is no deadline, only an incentive. Expect higher per-token rates than Flash-class models, offset in part by fewer tokens and faster finishes.
Paid Standard is $1.50 in and $7.50 out, flat across the full million-token window, with implicit caching at $0.075. Batch and Flex halve it to $0.75 and $3.75. Priority is $2.70 and $13.50. Search and Maps grounding each include 5,000 free monthly requests, then $14 per thousand. Treat these as indicative and check the console before you budget.
Take a week of real coding-agent sessions from your logs. Replay twenty on Hydrogen at high, twenty at xhigh, and twenty on Gemini 4 Argon. Score completion, tokens and wall-clock time. If xhigh closes most of the gap to Argon at a fraction of the cost, you have a routing rule.
Hydrogen is also the default model for the Antigravity agent in Gemini Managed Agents, and it is available in Android Studio, the Gemini app, Google Search, Gemini for business and Gemini in Google Sheets. Compare it with other coding models in the code generation directory.
Compare the model on the work you need to complete.
Gemini 4 Hydrogen is listed for Multimodal, Text generation, Coding & Code Generation. It accepts text, image, audio, video, pdf and produces text. Use the publisher’s input format and preprocessing for your chosen deployment.
Documented limits: 1.05M context. The listed route has starting rates of $1.50 input and $7.50 output per million tokens. Measure a completed task using the prompts, documents and tools your application will use. Include reasoning and retries in the result.
Use the model identifier and account requirements for the selected API route. Confirm its request limits and output format with one complete task before connecting a larger workflow.
These records share a directory category. Check the source card, output type and deployment requirements before treating a model as a replacement.
OpenAI
GPT-6 Cyber is OpenAI's restricted security model for authorised vulnerability investigation, reproduction and fix verification, for Daybreak Red customers.
particle-ai
Particle-1 is Particle's first software-engineering model, an RL-tuned DeepSeek V4.1 Flash that aims to solve coding tasks with fewer tokens.
Microsoft
MAI-Cyber-1.1-Flash is Microsoft AI's vulnerability-discovery model, available only inside its MDASH scanning harness to verified security customers.
recursive
Recursive Flint is an open-weight 39.8B MoE coding-agent model with 2.5B active parameters, free to try via OpenCode and OpenRouter.
Deepreinforce
Ornith-1.5 122B is an open MIT-licensed MoE coding and agent model with a 262K context, reasoning by default and single-80GB-GPU serving.
MiniMax
MiniMax M3.1 Flash is MiniMax's GA open-weight multimodal MoE for coding agents: 428B total, 23B active, 1M context, from $0.15 per 1M input tokens.