gemini-4-argon-high: benchmark scores, price and ranking
gemini-4-argon-high from Google ranks #10 of 743 LLMs on the LLMs Tiger aggregate benchmark score (59.5), based on 8 benchmark results captured up to 2026-10-13.
- Organisation
- Type
- Frontier model
- Open weights
- No
- Released
- 2026-09-30 (source: epoch)
- Cheapest API price
- $2.00 in / $10.00 out per million tokens (via gemini, observed 2026-10-13T19:47Z)
- Aggregate rank
- #10 of 743 (score 59.5)
Relative to other models, gemini-4-argon-high is strongest on SimpleQA Verified (#1 of 69) and weakest on FrontierSWE (#5 of 19).
Across the 8 benchmarks where at least ten models are scored, gemini-4-argon-high is in the top 10% on 4 and in the top half on 8.
Its cheapest listed input price, $2.00 per million tokens, is lower than that of 18% of the 427 models with a listed price.
gemini-4-argon-high benchmark results
| Benchmark | Score | Rank | Source | Captured |
|---|---|---|---|---|
| SimpleQA Verified | 76.0% | #1 of 69 | Epoch AI Benchmarking Hub | 2026-10-13 |
| SciCode | 61.8% | #3 of 106 | Epoch AI Benchmarking Hub | 2026-10-13 |
| LMArena Text Elo | 1,525 | #1 of 28 | LMArena | 2026-10-13 |
| CritPt | 27.1% | #11 of 115 | Epoch AI Benchmarking Hub | 2026-10-13 |
| WebDev Arena | 1,680 | #7 of 67 | Epoch AI Benchmarking Hub | 2026-10-13 |
| Furniture Assembly | 72.5% | #5 of 28 | Epoch AI Benchmarking Hub | 2026-10-13 |
| LMArena Code Elo | 1,678 | #7 of 30 | LMArena | 2026-10-13 |
| FrontierSWE | 55.0% | #5 of 19 | Epoch AI Benchmarking Hub | 2026-10-13 |
About gemini-4-argon-high
gemini-4-argon-high is a frontier model from Google. It was released on 2026-09-30, according to Epoch AI. The cheapest API listing we track charges $2.00 per million input tokens and $10.00 per million output tokens, through gemini. We hold 8 benchmark results for it across 3 areas, and 4 are in the top tenth of their boards.
Where gemini-4-argon-high stands
On our aggregate score, gemini-4-argon-high is #10 of 743 ranked models, in the top ten. Its score is 59.5; the scale is centred on 50, so it sits 9.5 points above the typical model. That beats 99% of the ranked field. Among the 29 ranked Google models it is #1. The next one down is gemini-3.1-pro-preview-high at 55.3.
gemini-4-argon-high, benchmark by benchmark
SimpleQA Verified is where gemini-4-argon-high does best: 76.0%, first of 69. The next result is GPT-6 Astra (max) at 75.6%, 0.4 points behind. It covers short factual questions. The board median is 41.0%, so gemini-4-argon-high is 35.0 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
On SciCode, gemini-4-argon-high scores 61.8% and ranks #3 of 106. Only 2 models score higher, led by Claude Opus 5.5 (max) at 66.9%, which is 5.1 points ahead. It covers scientific research code. The board median is 44.3%, so gemini-4-argon-high is 17.5 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
LMArena Text Elo is where gemini-4-argon-high does best: 1,525, first of 28. The next result is Claude Fable 5.1 (max) at 1,501, 24 points behind. It covers head-to-head chat preference (crowd votes). The board median is 1,478, so gemini-4-argon-high is 48 points above it. Source: LMArena, captured 2026-10-13.
On CritPt, gemini-4-argon-high scores 27.1% and ranks #11 of 115. Only 10 models score higher, led by GPT-5.6 Sol (max) at 32.3%, which is 5.1 points ahead. It covers unpublished research-level physics problems. The board median is 2.9%, so gemini-4-argon-high is 24.3 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
On WebDev Arena, gemini-4-argon-high reaches 1,680. It ranks #7 of 67, ahead of 90% of the field, and 141 points behind Claude Opus 5.5 (max) at 1,820. The results either side are Claude Opus 5 (max) (1,687) and Kimi K3 (max) (1,674). It covers building web apps, crowd-voted (Elo). The board median is 1,433, so gemini-4-argon-high is 246 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
gemini-4-argon-high manages 72.5% on Furniture Assembly, good for #5 of 28. Claude Opus 5.5 (max) leads at 83.3%, 10.8 points further up, and 82% of models sit below. The results either side are Claude Sonnet 5.5 (max) (75.0%) and Claude Fable 5.1 (max) (70.0%). It covers planning a physical assembly. The board median is 41.2%, so gemini-4-argon-high is 31.2 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
LMArena Code Elo puts gemini-4-argon-high at 1,678, #7 of 30. That beats 77% of the models tested, but 136 points short of Claude Opus 5.5 (max) (1,814). The results either side are GPT-6 Sol (max) (1,687) and Muse Spark 1.3 (max) (1,656). It covers crowd-voted coding answers. The board median is 1,622, so gemini-4-argon-high is 56 points above it. Source: LMArena, captured 2026-10-13.
On FrontierSWE, gemini-4-argon-high reaches 55.0%. It ranks #5 of 19, ahead of 74% of the field, and 10.5 points behind GPT-6 Astra (max) at 65.5%. The results either side are Claude Fable 5.1 (max) (56.3%) and Claude Opus 5 (max) (52.0%). It covers hard multi-hour software engineering. The board median is 29.5%, so gemini-4-argon-high is 25.5 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
Strengths and weaknesses
Software engineering: strong, ahead of 84% of the field on average across 4 benchmarks. Its best board here is SciCode (#3 of 106) and its weakest is FrontierSWE (#5 of 19). General knowledge and chat: strong, ahead of 92% of the field on average across 3 benchmarks. Its best board here is SimpleQA Verified (#1 of 69) and its weakest is Furniture Assembly (#5 of 28).
Price and value
The median input price across the 427 priced models is $0.500 per million tokens, so gemini-4-argon-high costs more than the typical listing. We found no cheaper model with a similar or better aggregate score, so gemini-4-argon-high is the lowest-priced option at its level. Similar scores cost more on Claude Opus 5 (max) ($5.00) and Claude Fable 5 (max) ($10.00).
Release history
The Google model released before it was Gemini 3.6 Flash (High) on 2026-07-21, which scores 48.9.
Reading these numbers
The 8 scores on this page come from 2 sources: Epoch AI Benchmarking Hub (6), LMArena (2). No official leaderboard supplied any of them, so treat them as third-party measurements. The leaderboard also lists 1 other reasoning-effort setting of this model, which redirect here. This page shows the highest.
gemini-4-argon-high is also listed at 1 other reasoning-effort setting; this page shows the highest-effort row, which is the one the leaderboard keeps by default.
Models ranked nearby
- #7 GPT-6 Sol (max) (60.0)
- #8 Claude Opus 5 (max) (60.0)
- #9 Claude Fable 5 (max) (59.9)
- #11 GPT-5.5 (xhigh) (58.0)
- #12 GPT-5.6 Terra (max) (57.8)
- #13 GPT-5.4 (xhigh) (56.9)
More from Google
- gemini-3.1-pro-preview-high (#16, 55.3)
- gemini-2.5-pro-preview-06-05 (#35, 53.5)
- Gemini 2.5 Pro Preview 05-06 (#48, 53.0)
- Gemini 2.5 Pro Preview 03-25 (#59, 52.6)
- Gemini 3 Deep Think (2/26) (#68, 52.4)
- gemini-3.5-flash-high (#74, 52.2)
Compare it on the live leaderboard → · How the aggregate score works