Gemini 4 Argon
Google's flagship reasoning model for complex software engineering, deep math and science, and long-horizon autonomous agents
Shogo Agent Index
Supported#6 of 9269.5range 62.9–76.2 · 32% of weight measured
Agentic coding
77
26% measured
Long-horizon autonomy
78
50% measured
Knowledge-work agents
59
52% measured
Computer use
61
estimated
Security & context learning
61
estimated
Price
- Input
- $2.00/M
- Output
- $10.00/M
- Cache read
- $0.10/M
- 3:1 blend
- $4.00/M
- Agent-effective
- $0.84/M
Cheapest host for agents: Google AI Studio at $0.42/M.
Agent-effective = 20 input tokens per output token, 85% cache hits. Why.
Agentic benchmarks
| Benchmark | Score | Normalised | Index weight | Harness / source |
|---|---|---|---|---|
| SWE-bench Verified | est. | 8% | predicted 88 | |
| Terminal-Bench 2.0 | est. | 8% | predicted 95 | |
| DeepSWE | est. | 6% | predicted 78 | |
| FrontierSWE | 55.0% | 5% | proximus · $129.36/task | |
| CursorBench | 52.9% | 4% | source · $4.57/task | |
| GSO-Bench | est. | 4% | predicted 71 | |
| METR Time Horizons | est. | 10% | predicted 85 | |
| Vending-Bench 2 | $13,718 | 6% | source | |
| PostTrainBench | 40.4% | 4% | OpenCode | |
| APEX-Agents | 82.4% | 8% | source | |
| GDPval | est. | 7% | predicted 68 | |
| Remote Labor Index | 17.1% | 5% | source | |
| DeepResearch Bench | est. | 5% | predicted 52 | |
| OSWorld 2.0 | est. | 6% | predicted 38 | |
| OSWorld | est. | 4% | predicted 94 | |
| Cybench | est. | 5% | predicted 93 | |
| CL-bench | est. | 5% | predicted 29 |
Other benchmarks
| Benchmark | Category | Score | Source |
|---|---|---|---|
| ARC-AGI | Reasoning | 98.5% | source |
| ARC-AGI-2 | Reasoning | 91.3% | source |
| Chess Puzzles | Reasoning | 59.0% | source |
| CritPt | Reasoning | 27.1% | source |
| DTBench | Knowledge | 97.6% | source |
| FrontierCode | Coding | 47.5% | devin-cli |
| FrontierMath Tiers 1-3 | Math | 86.0% | source |
| FrontierMath Tier 4 | Math | 78.0% | source |
| GPQA Diamond | Reasoning | 96.4% | source |
| LMCA | Knowledge | 63.7% | source |
| MirrorCode | Coding | 58.9% | source |
| OTIS Mock AIME | Math | 100.0% | source |
| ProofBench | Math | 99.0% | source |
| SciCode | Coding | 61.8% | source |
| SimpleBench | Reasoning | 91.7% | source |
| SimpleQA Verified | Knowledge | 76.0% | source |
| WebDev Arena | Coding | 1680 | source |
| WeirdML v3 | Coding | 20.3% | gemini_cli |
API hosts
Live prices and uptime for every host serving Gemini 4 Argon via OpenRouter.
| Host | Input | Output | Cache read | Agent $/M | Uptime 24h | Quant |
|---|---|---|---|---|---|---|
| Google AI Studio | $1.00 | $5.00 | $0.050 | $0.42 | 99.9% | — |
| $2.00 | $10.00 | $0.10 | $0.84 | 99.8% | — | |
| Google AI Studio | $2.00 | $10.00 | $0.10 | $0.84 | 96.3% | — |
| $3.60 | $18.00 | $0.18 | $1.52 | 95.2% | — | |
| Google AI Studio | $3.60 | $18.00 | $0.18 | $1.52 | 97.6% | — |