GGoogle · Released Oct 13, 2026
Gemini 4 Argon
Gemini 4 Argon is Google's flagship reasoning model, built for complex software engineering, deep mathematical and scientific problem solving, and autonomous multi-step agents. Google trained it to keep context across...
- #1APEX-Agents
- #1SimpleQA Verified
- #1GPQA Diamond
- #1ARC-AGI-1
- Released
- Oct 13 2026
- Context window
- 1M 944K out
- Input / 1M tokens
- $2.00 from $1.00
- Output / 1M tokens
- $10.00 from $5.00
Where to run it
2 providers compared
Cheapest: Google AI Studio at $1.00 / $5.00 per 1M input / output tokens — 72% less than the priciest option.
| Provider | Input | Output | Cache read | Context | Max output | Uptime | vs. best |
|---|---|---|---|---|---|---|---|
| Google AI Studio | $1.00 | $5.00 | $0.05 | 1M | 944K | 99.9%24h | Cheapest |
| Googleglobal | $2.00 | $10.00 | $0.10 | 1M | 944K | 99.4%24h | +100% |
| Google AI Studio | $2.00 | $10.00 | $0.10 | 1M | 944K | 94.7%24h | +100% |
| Googleus | $2.20 | $11.00 | $0.11 | 1M | 66K | 100%24h | +120% |
| Googleglobal/priority | $3.60 | $18.00 | $0.18 | 1M | 944K | 98.9%24h | +260% |
| Google AI Studiopriority | $3.60 | $18.00 | $0.18 | 1M | 944K | 96.7%24h | +260% |
Value
Price vs. performance
Gemini 4 Argon Other Google Other labs Value frontier · GPQA Diamond vs. input price per 1M tokens (log scale) · Drag to pan, pinch or Ctrl + scroll to zoom, double-click to zoom in.
Benchmarks
22 benchmark scores
| Benchmark | Score | Rank | Leader |
|---|---|---|---|
| FrontierSWESoftware engineering · Ultra-long-horizon engineering tasks with a 20-hour budget each | 55.0% | #5of 19 | GPT-6 Astra65.5% |
| CursorBenchSoftware engineering · Ambiguous multi-file tasks drawn from real Cursor sessions | 52.9% | #3of 16 | Claude Opus 5.557.8% |
| FrontierCodeSoftware engineering · Mergeable fixes for hard open-source issues | 47.5% | #11of 41 | Claude Opus 5.554.6% |
| SciCodeSoftware engineering · Research-coding problems from the natural sciences | 61.8% | #3of 93 | Claude Opus 5.566.9% |
| WeirdML v3Software engineering · Unfamiliar ML and data tasks with vague goals | 20.2% | #5of 22 | GPT-6 Astra42.2% |
| APEX-AgentsAgents & computer use · Long professional tasks in banking, consulting and law | 82.4% | #1of 51 | This model |
| Vending-Bench 2Agents & computer use · Run a simulated vending business for a year (final balance) | $13,718 | #3of 61 | GPT-6 Astra$15,515 |
| GDP.pdfWriting & content · Answer questions about real professional PDFs | 28.4% | #8of 38 | GPT-6.1 Sol35.0% |
| SimpleQA VerifiedWriting & content · Factual accuracy on 1,000 short questions | 76.0% | #1of 66 | This model |
| WebDev ArenaDesign & visual · People vote on which model builds the better website (Elo) | 1680 | #8of 88 | Claude Opus 5.51820 |
| Blueprint-Bench 2Design & visual · Turn apartment photos into 2D floor plans | 54.4% | #1of 32 | This model |
| Furniture AssemblyDesign & visual · Spot mistakes in photos of IKEA builds | 72.5% | #5of 34 | Claude Opus 5.583.3% |
| GPQA DiamondReasoning & knowledge · PhD-level biology, chemistry and physics questions | 96.4% | #1of 92 | This model |
| ARC-AGI-2Reasoning & knowledge · Few-shot abstract reasoning on grid puzzles | 91.3% | #5of 63 | GPT-6 Astra95.0% |
| ARC-AGI-1Reasoning & knowledge · Pattern induction from input–output examples | 98.5% | #1of 63 | This model |
| CritPtReasoning & knowledge · Open-ended, research-level physics problems | 27.1% | #13of 96 | GPT-5.6 Sol32.3% |
| Chess PuzzlesReasoning & knowledge · Find the single best move in 100 new puzzles | 59.0% | #9of 70 | GPT-6 Astra76.0% |
| LMCAReasoning & knowledge · Judge the quality of conceptual arguments like experts | 63.7% | #6of 96 | Claude Sonnet 5.568.3% |
| FrontierMathMaths · Expert-written problems, undergrad to early research (tiers 1–3) | 86.0% | #10of 72 | GPT-6.1 Sol93.7% |
| FrontierMath Tier 4Maths · Exceptionally hard research-level problems | 78.0% | #9of 62 | GPT-6.1 Sol100.0% |
| ProofBenchMaths · Graduate maths proofs verified in Lean 4 | 99.0% | #4of 70 | Claude Sonnet 5.5100.0% |
| OTIS Mock AIMEMaths · Competition-style problems harder than MATH Level 5 | 100.0% | #1of 90 | This model |
Specs
Details
- Model ID
google/gemini-4-argon- Developer
- Release date
- Oct 13, 2026 ·
- Context window
- 1,048,576 tokens
- Max output
- 943,718 tokens
- API price
- $2.00 / $10.00 per 1M input / output tokens
- Providers
- 2
- Input types
- audio, file, image, text, video
- Reasoning
- Yes
- Tool calling
- Yes
- Open weights
- No
- Free tier
- No
Release date, prices and providers from OpenRouter's catalog; buying directly from a provider can cost a little differently. Uptime is the share of OpenRouter requests each provider served over the last day. Benchmark scores are the best result per benchmark from Epoch AI, licensed CC BY 4.0. Rank counts every model tracked here with a score.