Text and chat · Google
Gemini
Google's Gemini models: Argon, the new flagship, and the fast Flash line. 1M context, and they take images, audio and video as input. Anonymized.
Anonymized High costWhat sets it apart
Gemini 4 Argon is Google's flagship reasoning model, scoring 53 on the index with a 1M window; below it, Gemini’s Flash line is the fast, low-priced end of the market: Gemini 3.8 Flash outputs over 230 tokens per second at $0.75 per million input tokens, with a 1M window that takes text, images, speech and video. Output is text only.
- Flash: over 230 tokens per second
- Takes speech and video as input
- 1M context on every version
Recommended for: ask about a photo, a screenshot or a scanned page ; work over a very long document or many notes at once.
Strengths and limits
Where it shines
- Reads images, audio and video, not only text
- 1M token context for long documents
- 3.8 Flash is built for coding and agents at a moderate price
- Argon reaches 53 on the index, 12 points above Gemini 3.8 Flash
Think twice when
- Anonymized, not private: Google's retention policy applies
- The Pro and 3 Flash models are previews and may change
- Argon is a slow starter: AA measures a time to first token of about 45 seconds at high effort
- 1M token context
- Input: text, image, audio, video
- Tool calling and JSON schema output
Use cases
Summarizing a recorded meeting
Listen to this meeting recording and list the decisions, who owns each action and any date mentioned.
Gemini models accept speech and video directly as input.
Fast answers at scale
Classify each of these 500 product reviews as positive, neutral or negative and give a one-line reason.
Flash-Lite is the cheapest tier, $0.30 per million input tokens.
Understanding a video
Watch this product demo and write a step-by-step help article with a screenshot description for each step.
Video input is supported across the Gemini versions in the catalog.
Against its rivals
- ClaudeClaude Sonnet 5.5 scores 56 against 53 for Gemini 4 Argon and 41 for Gemini 3.8 Flash.#1 of 90 on Artificial Analysis Intelligence Index (58 points)
- GPTGPT-6 Luna is cheaper still ($0.10 input) but only scores 38.#4 of 90 on Artificial Analysis Intelligence Index (53 points)
- GrokGrok 4.7 scores 46 but costs more and does not match Flash speed.#12 of 90 on Artificial Analysis Intelligence Index (46 points)
Version history
Each version, dated, with what it changed over the one before.
Hollow points are earlier versions no longer offered in Sub Rosa.
See the numbers
| Version | Released | points |
|---|---|---|
| Gemini 3 Flash Preview | 2025-12-17 | 18 |
| Gemini 3.1 Pro Preview | 2026-02 | 30 |
| Gemini 3.5 Flash | 2026-05 | 33 |
| Gemini 3.6 Flash | 2026-07 | 34 |
| Gemini 3.5 Flash-Lite | 2026-07 | 22 |
| Gemini 3.7 Flash | 2026-08 | 39 |
| Gemini 3.8 Flash | 2026-09 | 41 |
| Gemini 4 Argon | 2026-09-30 | 53 |
- Gemini 3 Flash Preview
- Introduced the Gemini 3 Flash tier: 1M context, text, image, speech and video input, $0.50 / $3 per million tokens.
- Gemini 3.1 Pro Preview
- Pro tier preview at $2 / $12 per million tokens, a reasoning model scoring 30 on the index.
- Gemini 3.5 Flash
- Reasoning Flash model scoring 33 on the index, priced at $1.50 / $9 per million tokens.
- Gemini 3.6 Flash
- Price drops to $0.75 / $3.75 per million tokens, with a score of 34 on the index.
- Gemini 3.5 Flash-Lite
- Cheapest tier at $0.30 / $2.50 per million tokens, over 320 tokens per second.
- Gemini 3.7 Flash
- Index score rises to 39 at the same $0.75 / $3.75 price.
- Gemini 3.8 Flash
- Index score of 41 at the same price; Google positions it for long-horizon software engineering and agents.
- Gemini 4 ArgonLatest
- Google's new flagship reasoning model: 53 on the index against 41 for Gemini 3.8 Flash, with a 1M window and long decode up to 1M output tokens.
- Takes text, images, video, audio and PDF; Google aims it at software engineering, science and multi-step agents.
Benchmarks
How to read these scores: what a benchmark tells you.
Artificial Analysis Intelligence Index: #5 of 90 models measured in the catalog. One score averaging about ten independent tests of reasoning, knowledge, math, coding and agent work, all run the same way for every model. Higher is better. Each model is listed at its highest reasoning effort. Two or three points apart is a close call; ten points is a different class of model.
See the numbers
| Model | points |
|---|---|
| 1. Claude Opus 5.5 | 58 |
| 2. Claude Sonnet 5.5 | 56 |
| 3. Claude Fable 5.1 | 53 |
| 4. GPT-6 Astra | 53 |
| 5. Gemini 4 Argon | 53 |
| 6. GPT-6.1 Sol | 52 |
| 7. Muse Wonder | 52 |
| 8. Claude Opus 5 | 51 |
| 9. Claude Fable 5 | 50 |
| 10. GPT-6 Sol | 48 |
| 22. Gemini 3.8 Flash | 41 |
| 26. Gemini 3.7 Flash | 39 |
All 8 models in this family
| Model | Use | Privacy | Price |
|---|---|---|---|
Gemini 3.8 Flashgemini-3-8-flash | Text | Anonymized | $2.108 per million tokens written$0.00148 a page written |
Gemini 3.7 Flashgemini-3-7-flash | Text | Anonymized | $2.108 per million tokens written$0.00148 a page written |
Gemini 3.6 Flashgemini-3-6-flash | Text | Anonymized | $2.108 per million tokens written$0.00148 a page written |
Gemini 3.5 Flashgemini-3-5-flash | Text | Anonymized | $4.249 per million tokens written$0.00297 a page written |
Gemini 3.5 Flash-Litegemini-3-5-flash-lite | Text | Anonymized | $1.405 per million tokens written$0.000984 a page written |
Gemini 3 Flash Previewgemini-3-flash-preview | Text | Anonymized | $1.686 per million tokens written$0.00118 a page written |
Gemini 3.1 Pro Previewgemini-3-1-pro-preview | Text | Anonymized | $6.744 per million tokens written$0.00472 a page written |
Gemini 4 Argongemini-4-argon | Text | Anonymized | $5.62 per million tokens written$0.00393 a page written |
Specifications
| Specification | Gemini 4 Argon | Gemini 3.8 Flash | Gemini 3.7 Flash | Gemini 3.5 Flash-Lite | Gemini 3.1 Pro Preview |
|---|---|---|---|---|---|
| Released | September 30, 2026 | September 2026 | August 2026 | July 2026 | February 2026 |
| Weights | · | Closed | Closed | Closed | Closed |
| Accepts | text, images, audio, video | text, images, audio, video | text, images, audio, video | text, images, audio, video | text, images, audio, video |
| Produces | text | text | text | text | text |
| Context | 1,000K tokens | 1,000K tokens | 1,000K tokens | 1,000K tokens | 1,000K tokens |
| Longest answer | 66K tokens | · | · | · | · |
| Abilities | Reads images, Uses Sub Rosa’s tools, Thinks before answering | Reads images, Uses Sub Rosa’s tools, Thinks before answering | Reads images, Uses Sub Rosa’s tools, Thinks before answering | Reads images, Uses Sub Rosa’s tools, Thinks before answering | Reads images, Uses Sub Rosa’s tools, Thinks before answering |
| Price | $5.62 per million tokens written | $2.108 per million tokens written | $2.108 per million tokens written | $1.405 per million tokens written | $6.744 per million tokens written |
| Privacy | Anonymized | Anonymized | Anonymized | Anonymized | Anonymized |
Prices, context and lengths from the live catalog on October 16, 2026; parameters and licenses from the maker.
Sources
Everything on this page comes from these pages, read on October 16, 2026 or as dated on each score.
Show the 13 sources
- artificialanalysis.ai/models/gemini-3-8-flash
- artificialanalysis.ai/models/gemini-3-7-flash
- artificialanalysis.ai/models/gemini-3-6-flash
- artificialanalysis.ai/models/gemini-3-5-flash
- artificialanalysis.ai/models/gemini-3-5-flash-lite
- artificialanalysis.ai/models/gemini-3-flash
- artificialanalysis.ai/models/gemini-3-1-pro-preview
- artificialanalysis.ai/models/gemini-4-argon
- ai.google.dev/gemini-api/docs/models
- artificialanalysis.ai/leaderboards/models
- datanorth.ai/news/google-launches-gemini-3-8-flash-gemini-3-8-flash-cyber
- deepmind.google/models/gemini/flash
- ai.google.dev/gemini-api/docs/models/gemini-4-argon