By the toolist.ai editors. Researched from public sources, not hands-on tested by toolist.ai. Updated October 13, 2026; prices checked October 13, 2026.
Gemini 4 Argon: price, context window and benchmarks
Gemini 4 Argon is the first Gemini 4 model, which Google calls its flagship reasoning model. It costs an introductory $2 per million input tokens and $10 per million output tokens through November 30, 2026, then $4 and $20 from December 1, and scores 52.6 on the Artificial Analysis Intelligence Index at high thinking.
Best for
- Long agentic coding and terminal work: 63.8 on the Artificial Analysis Coding Agent Index in the Antigravity CLI harness
- Deep math and science problems that need more than 65,536 tokens of thinking and output: with long decode (background execution), one response can use up to 1,048,576 tokens minus the input
- Long prompts while the introductory price lasts, with no surcharge above 200,000 tokens and cached input at $0.10 per million
Key facts
| Fact | Gemini 4 Argon | Source |
|---|---|---|
| Provider | Gemini API docs: Pricing | |
| API model name | gemini-4-argon | Gemini API docs: Gemini 4 Argon |
| Released | October 13, 2026 | Gemini API docs: Release notes |
| Input price | $2 per million tokens | Gemini API docs: Pricing |
| Cached input price | $0.10 per million tokens | Gemini API docs: Pricing |
| Output price | $10 per million tokens | Gemini API docs: Pricing |
| Context window | 1,048,576 tokens | Gemini API docs: Gemini 4 Argon |
| Maximum output | 65,536 tokens | Gemini API docs: Gemini 4 Argon |
| Knowledge cutoff | March 2026 | Google DeepMind: Gemini 4 Argon model card |
| Input and output | Text, images, video, audio and PDF in; text out | Gemini API docs: Gemini 4 Argon |
| Reasoning | Thinking is on by default. Thinking levels low, medium and high; medium is the API default | Gemini API docs: Gemini thinking |
| API and cloud platforms | Gemini API paid tier (in Google AI Studio and Android Studio), the Gemini platform on Google Cloud, and Google Antigravity | Google: Gemini 4 Argon: Built to solve your hardest problems |
| In the apps | Gemini app and AI Mode in Search on Google AI Pro and Ultra; Gemini for business | Google: Gemini 4 Argon: Built to solve your hardest problems |
| Data retention | On the paid tier, Google does not use prompts or responses to improve its products; it logs them for a limited period only to detect misuse and for legal disclosures | Gemini API Additional Terms of Service |
Introductory prices run through November 30, 2026; from December 1, 2026 input costs $4, output $20 and cached input $0.20 per million tokens. Output includes thinking tokens. Caching is implicit only; explicit caching is not supported. There is no long-context surcharge. Batch and Flex cost half; Priority costs 1.8 times the standard price. Grounding with Google Search and with Google Maps each has 5,000 free requests a month shared across Gemini 3 and newer models, then $14 per 1,000. Not available on the free tier.
What Gemini 4 Argon costs a month
Three typical workloads, priced at the standard API rates above. We calculated these; the provider does not publish them.
| Workload | Tokens a month | Cost a month |
|---|---|---|
| Support chatbot 100,000 replies, each 3,000 input and 400 output tokens, no caching | 300M in, 40M out | $1,000 |
| Document Q&A 5,000 questions over the same 60,000-token document set (90% cache hits), 800 output tokens each | 300M in, 4M out | $127 |
| Coding agent 200 tasks, each 1.5 million input tokens (90% cache hits) and 50,000 output tokens | 300M in, 10M out | $187 |
Reasoning models bill their thinking as output tokens, so heavy reasoning can raise the output count several times. Cache-write charges are not included.
Benchmarks
| Benchmark | Score | Source |
|---|---|---|
| Artificial Analysis Intelligence Index high thinking, the highest level | 52.6 | Artificial Analysis: Gemini 4 Argon (high) |
| Coding Agent Index Antigravity CLI harness, high thinking | 63.8 | Artificial Analysis: Coding agents |
| Terminal-Bench 4.0 high thinking | 57.1% | Artificial Analysis: Gemini 4 Argon (high) |
| Humanity's Last Exam high thinking | 57.1% | Artificial Analysis: Gemini 4 Argon (high) |
| GDPval-AA v2.1 (Elo) high thinking | 1624 | Artificial Analysis: Gemini 4 Argon (high) |
| SciCode high thinking | 61.8% | Artificial Analysis: Gemini 4 Argon (high) |
Benchmark and speed data marked as such come from Artificial Analysis.
What changed
Compared with Gemini 3.1 Pro Preview, Google's third-generation Pro model, which charges $2 input and $12 output per million tokens up to 200,000 prompt tokens and $4 and $18 above that, Argon's introductory $2 and $10 apply at any prompt length and cached input drops from $0.20 to $0.10. From December 1, 2026 Argon's $4 and $20 are above 3.1 Pro's prices for prompts up to 200,000 tokens. Argon supports implicit caching only; explicit caching is not supported. (Gemini API docs: Pricing)
Gemini 4 Argon compared
Other models
- Muse Wonder: $4 in, $20 out per million tokens
- Claude Haiku 5.5: $0.10 in, $0.50 out per million tokens
- GPT-6.1 Sol: $2 in, $10 out per million tokens
- Claude Sonnet 5.5: $2 in, $10 out per million tokens
- Claude Opus 5.5: $4 in, $20 out per million tokens
- GPT-6 Sol: $2 in, $10 out per million tokens
- GPT-6 Astra: $10 in, $50 out per million tokens
- Claude Fable 5.1: $10 in, $50 out per million tokens
- Claude Opus 5: $5 in, $25 out per million tokens
All prices side by side: AI model API price table. How we research: our method.
Sources
- Gemini API docs: Gemini 4 Argon (checked 2026-10-13)
- Gemini API docs: Pricing (checked 2026-10-13)
- Gemini API docs: Release notes (checked 2026-10-13)
- Gemini API docs: Gemini thinking (checked 2026-10-13)
- Google DeepMind: Gemini 4 Argon model card (checked 2026-10-13)
- Google: Gemini 4 Argon: Built to solve your hardest problems (checked 2026-10-13)
- Gemini API Additional Terms of Service (checked 2026-10-13)
- Artificial Analysis: Gemini 4 Argon (high) (checked 2026-10-13)
- Artificial Analysis: Coding agents (checked 2026-10-13)
- Gemini API docs: Long decode (checked 2026-10-13)
- OpenRouter: Gemini 4 Argon (checked 2026-10-13)