Google · released Oct 13, 2026
Gemini 4 Argon
Gemini 4 Argon is Google's new flagship and gives the best answers for everyday chat and research.
NewOur pick for everyday questionsOur pick for researchOur pick for professional work2 weeks oldThinks before answering
- Input price
- $2 / M
- Output price
- $10 / M
- Context
- 1M tokens
- Longest reply
- 33K
- Reads
- text, image, video, audio, pdf
- Knows up to
- 2026-03
Where it shines
- It tops the text arena at 1526 and multi-turn chat at 1547.
- It posts the best SimpleQA (76) and GPQA (96.4) scores.
- Google reports 77.9% on DeepSWE v1.1, ahead of Opus 5.5 at 74.2%.
Where it struggles
- Its API price doubles to $4 in, $20 out after Nov 30.
- Its web dev score of 1680 trails Opus 5.5 by 129 points.
Where you can use it
- Gemini appAppGoogle AI Pro ($20/mo) or Ultra ($100+/mo). Not on Free or AI Plus.
- AI Mode in SearchAppGoogle AI Pro and Ultra subscribers.
- Gemini APIAPI$2 in / $10 out until Nov 30, 2026, then $4 / $20. No free tier.
- OpenRouterAPIListed as
google/gemini-4-argon
How it scores on each job
Scores run 0 to 100 and blend every measure we have for that job. The number after the dot is its rank among the 33 models in the lineup.
The evidence
| Measure | Score | Rank in lineup | Source |
|---|---|---|---|
| Epoch Capabilities Index | 166 | #5 | Epoch AI |
| LMArena text | 1526 | #1 | LMArena |
| LMArena hard prompts | 1538 | #1 | LMArena |
| LMArena instruction following | 1528 | #2 | LMArena |
| LMArena multi-turn | 1547 | #1 | LMArena |
| LMArena creative writing | 1523 | #2 | LMArena |
| LMArena writing & language | 1524 | #2 | LMArena |
| LMArena coding | 1539 | #2 | LMArena |
| LMArena math | 1538 | #1 | LMArena |
| LMArena business & finance | 1512 | #1 | LMArena |
| LMArena legal & government | 1527 | #1 | LMArena |
| LMArena medicine | 1537 | #1 | LMArena |
| WebDev Arena | 1680 | #7 | LMArena |
| Agent Arena | #6 | #6 | LMArena |
| GPQA Diamond | 96% | #1 | Epoch AI |
| FrontierMath | 86% | #7 | Epoch AI |
| SimpleQA Verified | 76% | #1 | Epoch AI |
| ARC-AGI-2 | 91% | #4 | Epoch AI |
Compare it with these
4 weeks old
Anthropic
Claude Sonnet 5.5
Claude Sonnet 5.5 does most of what Opus does for half the price, especially agent work.
AgentsWebsites & appsCoding
- Price
- $2 in · $10 out
- Context
- 1M
4 weeks old
OpenAI
GPT-6.1 Sol
GPT-6.1 Sol matches Astra on hard math for a fifth of the price.
Hard problems
- Price
- $2 in · $10 out
- Context
- 1.1M
8 weeks old
Anthropic
Claude Fable 5.1
Claude Fable 5.1 reads long documents better than any other model and stays strong at most jobs.
Long documents
- Price
- $10 in · $50 out
- Context
- 1M