google/gemini-3-pro-preview| Provider | Input, $ per 1M | Output, $ per 1M | In our data since |
|---|---|---|---|
| qihang-ai | $0.57 | $3.43 | 31 Jul 2026 |
| Jiekou.AI | $1.8 | $10.8 | 31 Jul 2026 |
| gmi | $2 | $12 | 31 Jul 2026 |
| 302.AI | $2 | $12 | 31 Jul 2026 |
| Google DeepMind | $2 | $12 | 31 Jul 2026 |
| Google Vertex AI | $2 | $12 | 31 Jul 2026 |
| Helicone | $2 | $12 | 31 Jul 2026 |
| Merge Gateway | $2 | $12 | 31 Jul 2026 |
| OpenRouter | $2 | $12 | 31 Jul 2026 |
| Requesty | $2 | $12 | 31 Jul 2026 |
| OrcaRouter | $4 | $18 | 31 Jul 2026 |
Only the base tier and only per-token prices are shown. Batch, discounted and cached rates, as well as prices per image or per second of video, do not go into this table: they cannot stand in the same column as a price per million tokens.
The date in the last column is the day this price first entered our collection. The price may well be older: before that day we simply were not recording it. It has not changed since — otherwise a new row with a new date would stand in its place.
| Task set | Result | Run conditions | Measured by |
|---|---|---|---|
| Mock AIME 2024–2025 — olympiad problems | 91.38 % | · with a tuned harness | Epoch evaluations |
| GPQA Diamond — graduate-level questions | 90.15 % | · with a tuned harness | Epoch evaluations |
| GeoBench — locating a place from a photograph | 84 % | — | GeoBench leaderboard |
| ARC-AGI — generalising to unseen patterns | 75 % | · with a tuned harness | ARC Prize Leaderboard |
| SWE-bench Verified — fixing bugs in repositories | 72.93 % | · with a tuned harness | Epoch evaluations |
| SimpleQA Verified — factual accuracy | 72.9 % | — | Epoch evaluations |
| SimpleBench — trick questions | 71.68 % | — | SimpleBench Leaderboard |
| WeirdML — unusual machine learning tasks | 69.93 % | — | WeirdML Leaderboard |
| Terminal-Bench — working in the command line | 69.4 % | · with a tuned harness | https://www.tbench.ai/leaderboard/terminal-bench/2.0 |
| BALROG — game environments | 58.1 % | — | Balrog Leaderboard |
| GDPval — tasks from real occupations | 40.3 % | — | — |
| Humanity’s Last Exam — expert-level questions | 34.37 % | — | — |
| APEX-Agents | 31.5 % | — | — |
| ARC-AGI-2 | 31.11 % | — | — |
| Chess puzzles | 27.4 % | — | Epoch evaluations |
| GSO-Bench — code optimisation | 18.6 % | — | GSO Leaderboard |
| CritPt — physics problems | 6.9 % | — | — |
| Remote Labor Index — jobs from a freelance marketplace | 1.25 % | — | — |