Primary use stated or inferred cautiously from official documentation.
Gemini 4 Neon
Google multimodal model generally available since November 18. It combines coding, scientific reasoning and document work. It accepts text, images, audio, video and PDF and produces text, with a shared context window of approximately one million tokens.
The essential
Maximum input capacity when published by the source.
Documented maximum generation limit.
Verified availability channels.
Primary functional family and verified specialties.
Declared terms for API access, model weights, or self-hosting.
Limits and integration
| API ID | gemini-4-neon |
|---|---|
| Model type | Text and reasoning · Code · Multimodal |
| Access model | Paid proprietary |
| License | Proprietary |
| Deployment | Hosted API |
| Release | 18 NOV 2026 |
| Knowledge cutoff | Not published |
| Entrance | Text · Image · Video · Audio · PDF |
| Exit | Text |
| Context window | 1,048,576 tokens shared by input and output |
| maximum output | 65,536 synchronous tokens, up to 1,048,576 with long decode |
| Reasoning | low, medium, high and xhigh; high by default |
| Published tools | Features · Web search · Computer use |
| Structured Outings | Structured Outings |
| Batch processing | Available |
| Prompt cache | Implicit caching only; no explicit caching |
| Fine-tuning | Not published |
| Verified platforms | Gemini API · Google AI Studio |
Analyze long documents, reason about charts and carry out coding tasks with tools. Supports generateContent and the Interactions API. Office and OpenDocument processing allows selection of pages, sheets, cells or multimedia intervals.
The synchronous maximum is 65,536 tokens. Long output requires long decode and shares the window with input. Native image and audio generation are unsupported. Remove cached_content when migrating: only implicit caching is supported. Access and quotas depend on plan and region.
What it can do
Orientation
Analyze long documents, reason about charts and carry out coding tasks with tools. Supports generateContent and the Interactions API. Office and OpenDocument processing allows selection of pages, sheets, cells or multimedia intervals.
Context
1,048,576 tokens shared by input and output · 65,536 synchronous tokens, up to 1,048,576 with long decode
Tools and integration
Features · Web search · Computer use
Access
Gemini API · general availability
Documented cost
| Concept | Worth | Unit/condition |
|---|---|---|
| Standard input | USD 3 / 1M tokens | Standard API rate |
| Cached input | USD 0.15 / 1M tokens | Reading reused prefixes |
| Cache write or storage | Not published | The condition varies by provider |
| Standard output | USD 15 / 1M tokens | May include reasoning tokens |
| Batch input | USD 1.50 / 1M tokens | Asynchronous processing |
| Batch output | USD 7.50 / 1M tokens | Asynchronous processing |
Standard costs USD 3 input and USD 15 output per million tokens, including reasoning. Implicit cache reads cost USD 0.15 and require eligible prefixes from 4,096 tokens. Batch and Flex cost USD 1.50 and USD 7.50; Priority costs USD 5.40 and USD 27. Tools and searches may add charges. Neon has no free tier.
i Prices change and may depend on level, region or context length. Check the source before making a decision.
How to read the results
The card reports 65.2% on Terminal-bench 4.1 with cheating adjustment and 95.2% F1 on GraphWalks BFS from 256K to 1M tokens. These are Google results in its evaluation environments, not tests reproduced by Inferama. Context, effort or tool changes may affect results.
| Benchmark | Result | Metric | Source |
|---|---|---|---|
| Terminal-bench 4.1 | 65.2% | Cheating-adjusted accuracy | View source ↗ |
| GraphWalks BFS 256K–1M | 95.2% | F1 · BFS from 256K to 1M tokens | View source ↗ |
Inferama only highlights a “best result” when the metric, test set, configuration, and date allow for an equivalent comparison. The supplier's figures are presented as claims from its own source.
Chronology
gemini-4-neon
General availability; Batch, Flex and Priority are service modes, not new models.