Gemini 4
gemini-4-neon
Google Gemini 4 text reasoning series: Carbon is the most powerful in the series, aimed at complex, multi-step, long-horizon tasks; Argon offers a better balance of capability and cost. Supports text and image input, 1M token context
- Context window
- 1,048,576
- Input
- Max output
- 65,536 tokens
- Release date
- 2026-11-18
Try gemini-4-neon online
Pricing
gemini-4-neon /1M tokens
| Specifications | Unit Price |
|---|---|
| Input | $4.53515 |
| Cached input | $0.22676 |
| Output | $22.67574 |
gemini-4-neon @flex /1M tokens
| Specifications | Unit Price |
|---|---|
| Input | $2.26757 |
| Cached input | $0.11338 |
| Output | $11.33787 |
Overview
Positioning and overview
Gemini 4 Neon is a text output model in the Google Gemini 4 series positioned between Hydrogen and Carbon, balancing reasoning depth against completion speed. Google positions it for complex software engineering, analytical reasoning, enterprise knowledge work, and scientific research workflows—tasks that need both fairly deep reasoning and fairly fast completion. On HiAPI it is called through the Chat Completions API, returns text results, and supports streaming output.
Key highlights
- 1M-token context—long documents, codebase explanations, and multi-turn conversations can fit in a single task, reducing the need to split and reorganize them.
- Up to 65,536 output tokens per request—long analyses, complete code files, or structured reports can be returned in one pass.
- Three thinking levels (low / medium / high)—trade off response speed against reasoning depth according to task difficulty.
- Two tiers, standard and @flex—the same model; the @flex price is about half the standard tier but has higher latency, making it suitable for batch and offline tasks.
Capabilities
Complex coding and troubleshooting: it can read multiple source files, error logs, and requirement notes at once and produce cross-file change plans with explanations of the reasoning, making it suitable for multi-step engineering tasks.
Analytical reasoning: it performs multi-step reasoning, comparison, and synthesis over long materials, organizing conclusions into bullet points, tables, or report-structured text.
Long-context knowledge work: backed by a 1M-token window, it can hold an entire contract, technical document, or meeting transcript for Q&A, drafting, and rewriting.
Thinking-level adjustment: use reasoning_effort to select low / medium / high—use the low level for simple questions to gain speed, and the high level for hard problems to gain accuracy.
Use cases
- Developers handling cross-file changes—put the relevant source code and error logs into the context, raise the thinking level, and get a change plan with explanations.
- Analysis and research teams synthesizing long materials—input reports, papers, or research notes and output structured key points and comparisons of conclusions.
- Enterprise knowledge base Q&A and document drafting—answer questions based on internal long documents, or generate documentation and first drafts of proposals.
- Latency-insensitive batch tasks—route large-batch summarization, classification, and data organization through the @flex tier at about half the standard cost.
Limitations and selection advice
Text output only. When called through HiAPI, the thinking level offers three options: low / medium / high. The @flex tier has higher latency than the standard tier and may be rejected first when upstream capacity is tight, so it is not suitable for real-time conversation. If speed and cost matter more, choose Gemini 4 Hydrogen; for the most complex, longest-running tasks, consider Gemini 4 Carbon.
What's new
View changelog →- New model gemini-4-neon is live: the Gemini 4 model balancing reasoning depth and speed
- New model gemini-4-hydrogen launches: the speed and cost tier of the Gemini 4 series
- New model New Models Available: Gemini 4 Carbon, Gemini 4 Argon, Gemini 3.8 Flash
Articles
View all articles →-
Guide
Getting Started with gemini-4-neon: Text Capability Boundaries, Integration Parameters, and Cost Control
Integration Decisions and Practical Points for Developers
-
Comparison
How to choose between gemini-4-neon and similar models: a developer's cross-comparison and migration guide
From capability boundaries and token pricing to migration cost, a practical model-selection decision framework
-
Tutorial
gemini-4-neon Real-World Use Cases in Practice: 5 High-Value Applications and Integration Essentials
Related Models
View all models →Same Type
-
Kimi K3.1 Pro Moonshot AI $0.216 /1M tokens -
Grok 4 xAI $0.210 /1M tokens -
Claude 5.5 Anthropic $0.100 /1M tokens -
Kimi K3 Moonshot AI $0.232 /1M tokens -
GPT 6 OpenAI $0.0082 /1M tokens -
DeepSeek V4 DeepSeek $0.014 /1M tokens -
Gemini 3.8 Google $0.227 /1M tokens -
GPT 5.6 OpenAI $0.050 /1M tokens -
Gemini 4 Google $0.057 /1M tokens -
GLM 5.3 Zhipu AI $0.855 /1M tokens
Same Provider
-
Veo 3.1 Lite Google $0.060 /second -
Nano Banana Google $0.011 /image -
Nano Banana 2 Google $0.018 /image -
Gemini Omni 1.2 Google $0.023 /second -
Gemini 4 Google $0.057 /1M tokens -
Lyria 3 Pro Google $0.140 /song -
Imagen 4 Google $0.030 /image -
Gemini 3.8 Google $0.227 /1M tokens -
Veo 3.1 Fast Google $0.170 /second -
Google Virtual Try-On Google $0.107 /image -
Veo 3.1 Google $0.290 /second -
Nano Banana 2 Lite Google $0.033 /image