Gemini 4 Argon vs GPT-6 Astra
Published test results, prices, limits and features, side by side. Share the link to show someone exactly this comparison.
Gemini 4 ArgonGoogle
GPT-6 AstraOpenAI
At a glance
Capability index
- Gemini 4 Argon164.6
- GPT-6 Astra166.4
GPT-6 Astra is 1.8 points higher.
Price per 1M tokens
- Gemini 4 Argon$4.00
- GPT-6 Astra$20.00
Gemini 4 Argon is 5 times cheaper.
Context window
- Gemini 4 Argon1M tokens
- GPT-6 Astra1.1M tokens
GPT-6 Astra takes in 0% more.
Released
- Gemini 4 ArgonOct 13, 2026
- GPT-6 AstraSep 4, 2026
Gemini 4 Argon is 39 days newer.
Where each scores higher
On the 13 tests both have taken, each on its own scale. A gap of a point or two can sit within a test’s margin of error.
Gemini 4 Argon
- Arena text+48 points
- Arena coding+17 points
- Arena math+47 points
- SimpleQA Verified+0.4 points
- Arena creative writing+67 points
- Arena instruction following+56 points
- GPQA Diamond+0.6 points
- Arena hard prompts+48 points
GPT-6 Astra
- Arena WebDev+102 points
- FrontierMath Tiers 1–3+7.7 points
- FrontierMath Tier 4+19.5 points
- Chess puzzles+15.0 points
Level on Mock AIME 2024–2025.
10 more tests have results for only one of them; the table below lists every result.
Everything side by side
| Measure | Gemini 4 ArgonGoogle | GPT-6 AstraOpenAI |
|---|---|---|
| Model | ||
| Maker | OpenAI | |
| Sold here by | OpenAI | |
| API model ID | gemini-4-argon | gpt-6-astra |
| Released | Oct 13, 2026 | Sep 4, 2026 |
| Knowledge cutoff | Mar 2026 | Apr 30, 2026 |
| Weights | Closed: hosted access only | Closed: hosted access only |
| Developer’s country | United States | United States |
| Hosts selling it | 24including Google directly | 35including OpenAI directly |
| Training compute | Not published | 9.4 × 10²⁶ FLOPEpoch AI estimate |
| Test results | ||
| Capability index | 164.6#5 of 277 | 166.4 (best of these models)#2 of 277 |
| Arena text | 1525 (best of these models)#1 of 415 | 1477#30 of 415 |
| LiveBench overall | Not tested | 82.2%#4 of 70 |
| Arena WebDev | 1678#9 of 142 | 1780 (best of these models)#2 of 142 |
| Arena coding | 1561 (best of these models)#1 of 410 | 1543#9 of 410 |
| LiveBench coding | Not tested | 80.4%#21 of 70 |
| Mock AIME 2024–2025 | 100.0%#1 of 303 | 100.0%#1 of 303 |
| FrontierMath Tiers 1–3 | 86.0%#10 of 117 | 93.7% (best of these models)#1 of 117 |
| FrontierMath Tier 4 | 78.0%#15 of 73 | 97.6% (best of these models)#2 of 73 |
| Arena math | 1529 (best of these models)#1 of 400 | 1482#34 of 400 |
| LiveBench mathematics | Not tested | 96.8%#4 of 70 |
| SimpleQA Verified | 76.0% (best of these models)#1 of 90 | 75.6%#2 of 90 |
| CursorBench | 52.9%#6 of 71 | Not tested |
| LiveBench agentic coding | Not tested | 57.3%#22 of 70 |
| DeepSWE | Not tested | 74.1%#1 of 69 |
| Arena creative writing | 1516 (best of these models)#2 of 413 | 1450#39 of 413 |
| Arena instruction following | 1528 (best of these models)#1 of 415 | 1472#35 of 415 |
| LiveBench language | Not tested | 89.4%#4 of 70 |
| LiveBench instructions | Not tested | 75.6%#8 of 70 |
| LiveBench data analysis | Not tested | 83.0%#1 of 70 |
| GPQA Diamond | 96.4% (best of these models)#1 of 323 | 95.8%#2 of 323 |
| Chess puzzles | 59.0%#9 of 243 | 74.0% (best of these models)#1 of 243 |
| Arena hard prompts | 1550 (best of these models)#1 of 415 | 1502#30 of 415 |
| LiveBench reasoning | Not tested | 92.7%#1 of 70 |
| Price per million tokens | ||
| Input | $2.00 (best of these models) | $10.00 |
| Output | $10.00 (best of these models) | $50.00 |
| Blended, 3 input to 1 output | $4.00 (best of these models) | $20.00 |
| Cached input | $0.10 (best of these models) | $1.00 |
| Writing to the cache | Not reported | $12.50 |
| Audio input | $2.00 | Not reported |
| Long requests | No separate price reported | $20.00 in, $75.00 outabove 272K tokens |
| The maker’s own price | This listing | This listing |
| Limits | ||
| Context window | 1M tokens | 1.1M tokens (best of these models) |
| Max input | 1M tokens (best of these models) | 922K tokens |
| Max output | 32.8K tokens | 128K tokens (best of these models) |
| Features | ||
| Reads | audio, images, PDFs, text and video | images, PDFs and text |
| Produces | text | text |
| Reasoning | Yeseffort low, medium, high | Yeseffort low, medium, high, xhigh, max |
| Tool calling | Yes | Yes |
| Structured output | Yes, such as JSON | Yes, such as JSON |
| File attachments | Yes | Yes |
| Temperature setting | Supported | Fixed by the provider |
Bold marks the better value in each row. Prices are each model’s own list price, or the middle price across hosts when the maker does not sell it, blended as three input tokens for every output token. Test results as published by LiveBench (Apache-2.0), Epoch AI (CC BY 4.0), Datacurve DeepSWE, compiled by Epoch AI (CC BY 4.0), Cursor CursorBench, compiled by Epoch AI (CC BY 4.0), Arena (CC BY 4.0); model details and prices from models.dev (MIT). Confirm pricing with the provider before use.