Gemini 3 Pro Preview
| Released | 2025-11-18 |
|---|---|
| Price per 1M | $2.00 in · $12.00 out |
| Overall score | 56.0 · rank 67 |
| Context window | 1,048,576 tokens |
| Provider | |
| Weights | Proprietary |
Scores
| Dimension | Score | Rank | Benchmarks |
|---|---|---|---|
| Overall: Balanced | 56.0 | 67 | 38 |
| Overall: Coding | 53.7 | 79 | 24 |
| Overall: Reasoning | 57.9 | 57 | 15 |
| Overall: Agentic work | 55.0 | 78 | 22 |
| Overall: Frontend | 49.1 | 97 | 9 |
| Coding | 55.4 | 67 | 3 |
| Agents | 53.9 | 81 | 11 |
| Reasoning | 59.0 | 54 | 8 |
| Math | 55.8 | 61 | 5 |
| Knowledge | 58.9 | 34 | 2 |
| Frontend | 43.7 | 98 | 2 |
| General | 56.6 | 51 | 3 |
| Multimodal | 63.3 | 2 | 4 |
| Value vs price | +3.7 | 91 |
Configurations
| Configuration | Balanced | Benchmarks |
|---|---|---|
| Unspecified setting | 56.2 ±1.1 | 30 |
| High | 54.9 ±2.4 | 9 |
| Low | 50.6 ±7.9 | 1 |
Evidence
Coding 55.4 ±2.9
| Benchmark | Result | Source | Date |
|---|---|---|---|
| GSO | 18.6% | Epoch AI | 2026-11-11 |
| LiveCodeBench (Vals run) | 86.4% | Vals AI | 2026-09-01 |
| SWE-bench Verified (Epoch run) | 72.9% | Epoch AI | 2026-02-13 |
| SWE-bench Verified (Vals run) | 76.4% | Vals AI | 2026-09-01 |
Agents 53.9 ±1.6
| Benchmark | Result | Source | Date |
|---|---|---|---|
| Terminal-Bench 2.0 | 69.4% | Epoch AI | 2025-11-18 |
| Remote Labor Index | 1.3% | Epoch AI | 2026-11-11 |
| METR 50% Time Horizon | $224 | METR | 2026-11-11 |
| tau2-bench Banking Knowledge | 18.0% | tau2-bench | 2026-03-02 |
| BFCL v4 | 72.5% | BFCL | 2026-04-12 |
| METR task suite | 71.0% | Epoch AI | 2026-11-11 |
| tau2-bench Airline | 80.5% | tau2-bench | 2026-03-02 |
| DeepResearch Bench | 46.3% | Epoch AI | 2026-11-11 |
| BALROG | 58.1% | Epoch AI | 2026-11-11 |
| Vending-Bench 2 | $5,478 | Epoch AI | 2026-11-11 |
| tau2-bench Telecom | 91.0% | tau2-bench | 2026-03-02 |
| tau2-bench Retail | 75.9% | tau2-bench | 2026-03-02 |
Reasoning 59.0 ±2.1
| Benchmark | Result | Source | Date |
|---|---|---|---|
| ARC-AGI-2 | 31.1% | Epoch AI | 2026-11-11 |
| Extended NYT Connections | 94.4% | Lech Mazur | 2026-11-10 |
| ARC-AGI-2 (ARC Prize verified) | 31.1% | ARC Prize | 2026-11-11 |
| CritPt | 6.9% | Epoch AI | 2026-11-11 |
| ARC-AGI-1 (ARC Prize verified) | 75.0% | ARC Prize | 2026-11-11 |
| ARC-AGI-1 | 75.0% | Epoch AI | 2026-11-11 |
| EnigmaEval | 18.2% | Epoch AI | 2026-11-11 |
| GPQA Diamond | 92.6% | Epoch AI | 2025-11-19 |
| GPQA Diamond (Vals run) | 91.7% | Vals AI | 2026-09-01 |
| Kagi Reasoning Benchmark | 80.1% | Kagi | 2026-11-11 |
| SimpleBench | 76.4% | SimpleBench | 2025-11-18 |
| SimpleBench | 76.4% | Epoch AI | 2026-11-11 |
Math 55.8 ±1.8
| Benchmark | Result | Source | Date |
|---|---|---|---|
| FrontierMath Tiers 1-3 (2025 set) | 37.6% | Epoch AI | 2025-11-21 |
| ProofBench | 20.0% | Epoch AI | 2026-11-11 |
| FrontierMath Tier 4 (2025 set) | 18.8% | Epoch AI | 2025-11-21 |
| OTIS Mock AIME 2024-2025 | 91.4% | Epoch AI | 2025-11-19 |
| MathArena Final-Answer Competitions | 67.0% | MathArena | 2026-11-11 |
Knowledge 58.9 ±1.3
| Benchmark | Result | Source | Date |
|---|---|---|---|
| Humanity's Last Exam | 37.5% | Epoch AI | 2026-11-11 |
| MMLU-Pro (Vals run) | 90.1% | Vals AI | 2026-09-01 |
Frontend 43.7 ±0.7
| Benchmark | Result | Source | Date |
|---|---|---|---|
| Vibe Code Bench v1.1 | 14.3% | Vals AI | 2026-11-09 |
| LMArena WebDev | 1,441 | LMArena | 2026-11-09 |
General 56.6 ±3.8
| Benchmark | Result | Source | Date |
|---|---|---|---|
| LMArena Text | 1,486 | LMArena | 2026-11-09 |
| EQ-Bench Creative Writing v3 | 1,524 | EQ-Bench | 2026-11-11 |
| EQ-Bench Longform Creative Writing | 68.8% | EQ-Bench | 2026-11-11 |
Multimodal 63.3 ±4.7
| Benchmark | Result | Source | Date |
|---|---|---|---|
| LMArena Vision | 1,289 | LMArena | 2026-11-09 |
| MMMU-Pro (Vals run) | 87.5% | Vals AI | 2026-09-01 |
| VPCT | 91.0% | Epoch AI | 2026-11-11 |
| GeoBench | 84.0% | Epoch AI | 2026-11-11 |
Results not used in the score
| Benchmark | Result | Source | Why not used |
|---|---|---|---|
| BFCL v4 | 68.1% | BFCL | A stronger configuration result was used |
| Terminal-Bench 2.0 | 65.2% | Epoch AI | A stronger configuration result was used |
| Terminal-Bench 2.0 | 62.3% | Epoch AI | A stronger configuration result was used |
| Terminal-Bench 2.0 | 62.2% | Epoch AI | A stronger configuration result was used |
| Terminal-Bench 2.0 | 61.8% | Epoch AI | A stronger configuration result was used |
| Terminal-Bench 2.0 | 61.1% | Epoch AI | A stronger configuration result was used |
| Terminal-Bench 2.0 | 56.9% | Epoch AI | A stronger configuration result was used |
| Terminal-Bench 2.0 | 56.0% | Epoch AI | A stronger configuration result was used |
Names matched to this model
gemini-3-pro, gemini-3-pro-preview