Gemini 3.1 Pro Preview
Reasoning-first Gemini preview for agentic coding and complex problem solving
Shogo Agent Index
Supported#45 of 9246.0range 41.6–50.5 · 64% of weight measured
Agentic coding
47
74% measured
Long-horizon autonomy
59
100% measured
Knowledge-work agents
36
52% measured
Computer use
39
estimated
Security & context learning
49
50% measured
Agentic benchmarks
| Benchmark | Score | Normalised | Index weight | Harness / source |
|---|---|---|---|---|
| SWE-bench Verified | 75.6% | 8% | source | |
| Terminal-Bench 2.0 | 80.2% | 8% | TongAgents | |
| DeepSWE | 11.7% | 6% | mini-swe-agent · $2.14/task | |
| FrontierSWE | est. | 5% | predicted 19 | |
| CursorBench | est. | 4% | predicted 39 | |
| GSO-Bench | 22.6% | 4% | OpenHands | |
| METR Time Horizons | 77.0% | 10% | source | |
| Vending-Bench 2 | $3,774 | 6% | source | |
| PostTrainBench | 18.1% | 4% | OpenCode | |
| APEX-Agents | 35.1% | 8% | source | |
| GDPval | est. | 7% | predicted 50 | |
| Remote Labor Index | est. | 5% | predicted 5 | |
| DeepResearch Bench | 47.8% | 5% | source | |
| OSWorld 2.0 | est. | 6% | predicted 12 | |
| OSWorld | est. | 4% | predicted 79 | |
| Cybench | est. | 5% | predicted 78 | |
| CL-bench | 20.8% | 5% | source |
Other benchmarks
| Benchmark | Category | Score | Source |
|---|---|---|---|
| AlgoTune | Coding | 2.02× | source |
| ARC-AGI | Reasoning | 98.0% | source |
| ARC-AGI-2 | Reasoning | 77.1% | source |
| BALROG | Long-horizon | 57.0% | source |
| Chess Puzzles | Reasoning | 51.0% | source |
| CritPt | Reasoning | 17.7% | source |
| DTBench | Knowledge | 97.1% | source |
| ExploitBench | Security & context | 26.1% | source |
| FrontierMath Tiers 1-3 | Math | 59.6% | source |
| FrontierMath Tier 4 | Math | 26.8% | source |
| GeoBench | Multimodal | 84.0% | source |
| GPQA Diamond | Reasoning | 94.4% | source |
| Humanity's Last Exam | Reasoning | 46.4% | source |
| LMCA | Knowledge | 53.8% | source |
| MirrorCode | Coding | 8.9% | source |
| OTIS Mock AIME | Math | 95.6% | source |
| ProofBench | Math | 26.0% | source |
| SciCode | Coding | 58.7% | source |
| SimpleBench | Reasoning | 79.6% | source |
| SimpleQA Verified | Knowledge | 73.5% | source |
| WebDev Arena | Coding | 1447 | source |
API hosts
Live prices and uptime for every host serving Gemini 3.1 Pro Preview via OpenRouter.
| Host | Input | Output | Cache read | Agent $/M | Uptime 24h | Quant |
|---|---|---|---|---|---|---|
| $1.00 | $6.00 | $0.10 | $0.51 | 0.0% | — | |
| Google AI Studio | $1.00 | $6.00 | $0.10 | $0.51 | 99.9% | — |
| $2.00 | $12.00 | $0.20 | $1.02 | 100.0% | — | |
| Google AI Studio | $2.00 | $12.00 | $0.20 | $1.02 | 99.1% | — |
| $3.60 | $21.60 | $0.36 | $1.83 | 100.0% | — | |
| Google AI Studio | $3.60 | $21.60 | $0.36 | $1.83 | — | — |