Google DeepMind

Gemini 3.1 Pro Preview

Type
language model
Context
1,049K tokens
Max output
66K
Released
19 February 2026
Knowledge cutoff
February 2026
API string
gemini/gemini-3.1-pro-preview

Prices by provider

Provider Input, $ per 1M Output, $ per 1M In our data since
vivgrid $2 $12 31 Jul 2026
daoxe $2 $12 31 Jul 2026
frogbot $2 $12 31 Jul 2026
auriko $2 $12 31 Jul 2026
ofox $2 $12 31 Jul 2026
perplexity-agent $2 $12 31 Jul 2026
AIHubMix $2 $12 31 Jul 2026
Abacus.AI $2 $12 31 Jul 2026
CrossModel $2 $12 31 Jul 2026
FastRouter $2 $12 31 Jul 2026
GitHub Copilot $2 $12 31 Jul 2026
Google DeepMind $2 $12 31 Jul 2026
Google Vertex AI $2 $12 31 Jul 2026
Kilo Code $2 $12 31 Jul 2026
LLM Gateway $2 $12 31 Jul 2026
Merge Gateway $2 $12 31 Jul 2026
NanoGPT $2 $12 31 Jul 2026
OpenRouter $2 $12 31 Jul 2026
Vercel $2 $12 31 Jul 2026
ZenMux $2 $12 31 Jul 2026
Venice AI $2.5 $15 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.

Measurement results

Task set Result Run conditions Measured by
Arena Score in Russian 1,495.77 1,488–1,503
Arena Score, mathematics 1,489.07 1,480–1,498
Arena Score in French 1,488.01 1,474–1,502
Arena Score, expert questions 1,486.42 1,479–1,494
Arena Score, hard prompts 1,484.41 1,480–1,489
Arena Score, programming 1,483.89 1,479–1,489
Arena Score, multi-turn dialogue 1,482.41 1,476–1,489
Arena Score, long queries 1,482.3 1,477–1,487
Arena Score, creative writing 1,481.61 1,475–1,488
Arena Score in Spanish 1,479.93 1,467–1,493
Arena Score, overall 1,479.46 1,476–1,483
Arena Score in English 1,478.03 1,474–1,483
Arena Score, instruction following 1,467.43 1,462–1,473
Arena Score, web development 1,446.83 1,441–1,452
Arena Score, understanding diagrams 1,312.53 1,305–1,320
Arena Score, text recognition in images 1,305.79 1,300–1,311
Arena Score, working with images 1,295.15 1,289–1,301
ARC-AGI — generalising to unseen patterns 98 % · with a tuned harness ARC Prize Leaderboard
Mock AIME 2024–2025 — olympiad problems 95.6 % · with a tuned harness Epoch evaluations
GPQA Diamond — graduate-level questions 92.13 % · with a tuned harness Epoch evaluations
Terminal-Bench — working in the command line 80.2 % · with a tuned harness https://www.tbench.ai/leaderboard/terminal-bench/2.0
SimpleQA Verified — factual accuracy 77.3 % Epoch evaluations
ARC-AGI-2 77.1 %
SimpleBench — trick questions 75.52 % SimpleBench Leaderboard
WeirdML — unusual machine learning tasks 72.1 % WeirdML Leaderboard
FrontierMath, levels 1–3 59.65 % Epoch evaluations
BALROG — game environments 57 % Balrog Leaderboard
Chess puzzles 52.65 % Epoch evaluations
Humanity’s Last Exam — expert-level questions 43.74 %
APEX-Agents 33.5 %
FrontierMath, level 4 — research-grade problems 26.83 % Epoch evaluations
GSO-Bench — code optimisation 22.55 % https://gso-bench.github.io/index.html
CritPt — physics problems 17.71 %
Arena Score, agent tasks -0.01 0–0