Compare / head-to-head
Gemini 3.1 Pro Previewvs
Grok 4.5
Gemini 3.1 Pro Preview leads 16 of 26 shared benchmarks. Grok 4.5 is 1.8x cheaper per token. Gemini 3.1 Pro Preview has the larger context window (1.0M tokens).
Benchmarks from cited public sources; pricing from official pages; status from official provider feeds.
Shared benchmarks
16 – 10
Gemini 3.1 Pro Preview leads
Cheaper per token
Grok 4.5
1.8x cheaper, input + output
Larger context
Gemini 3.1 Pro Preview
1.0M tokens
Providers
2 providers
Google · xAI
Gemini 3.1 Pro Preview
Google · released 2026-02-19
textvisionaudio
Quality
Benchmark matrix
26 shared · 10 only Gemini 3.1 Pro Preview · 1 only Grok 4.5| Benchmark | Gemini 3.1 Pro Preview | Grok 4.5 | Δ | Edge |
|---|---|---|---|---|
| Reported by both · 26 | ||||
| GPQA Diamond | 94.3% ↗ | 93.4% ↗epoch run | +0.9 pt | Gemini 3.1 Pro Preview |
| LMArena Elo | 1480.1 ↗ | 1450.1 ↗ | +30 | Gemini 3.1 Pro Preview |
| APEX-Agents (Mercor) | 35.3% ↗ | 56.2% ↗ | -20.9 pt | Grok 4.5 |
| Chess Puzzles (Epoch AI run) | 55% ↗epoch run | 36% ↗epoch run | +19 pt | Gemini 3.1 Pro Preview |
| DeepSWE v1.1 (Datacurve) | 11.7% ↗ | 53.8% ↗ | -42.1 pt | Grok 4.5 |
| FrontierMath Tier 4 v2 (Epoch AI run) | 26.8% ↗epoch run | 24.4% ↗epoch run | +2.4 pt | Gemini 3.1 Pro Preview |
| FrontierMath Tiers 1-3 v2 (Epoch AI run) | 59.6% ↗epoch run | 57.2% ↗epoch run | +2.4 pt | Gemini 3.1 Pro Preview |
| Furniture Assembly (Epoch AI run) | 26.7% ↗epoch run | 22.5% ↗epoch run | +4.2 pt | Gemini 3.1 Pro Preview |
| LiveBench Agentic Coding (LiveBench) | 44.1% ↗ | 56.5% ↗ | -12.4 pt | Grok 4.5 |
| LiveBench Coding (LiveBench) | 76.5% ↗ | 68.6% ↗ | +7.9 pt | Gemini 3.1 Pro Preview |
| LiveBench Data Analysis (LiveBench) | 78.5% ↗ | 73% ↗ | +5.5 pt | Gemini 3.1 Pro Preview |
| LiveBench Instruction Following (LiveBench) | 79.1% ↗ | 71.5% ↗ | +7.6 pt | Gemini 3.1 Pro Preview |
| LiveBench Language (LiveBench) | 85.4% ↗ | 82.8% ↗ | +2.6 pt | Gemini 3.1 Pro Preview |
| LiveBench Mathematics (LiveBench) | 91% ↗ | 90.8% ↗ | +0.2 pt | Gemini 3.1 Pro Preview |
| LiveBench Reasoning (LiveBench) | 84% ↗ | 87.2% ↗ | -3.2 pt | Grok 4.5 |
| LMArena Agent (LMArena) | -0.0771 ↗ | 0.0121 ↗ | -0.1 | Grok 4.5 |
| LMArena Vision (LMArena) | 1279.3 ↗ | 1279 ↗ | +0.3 | Gemini 3.1 Pro Preview |
| LMArena WebDev (LMArena) | 1446.2 ↗ | 1551.8 ↗ | -105.6 | Grok 4.5 |
| OTIS Mock AIME 2024-2025 (Epoch AI run) | 95.6% ↗epoch run | 97.8% ↗epoch run | -2.2 pt | Grok 4.5 |
| SAGE (Vals AI) | 48.7% ↗ | 35% ↗ | +13.7 pt | Gemini 3.1 Pro Preview |
| SimpleBench (SimpleBench) | 79.6% ↗ | 70% ↗ | +9.6 pt | Gemini 3.1 Pro Preview |
| SimpleQA Verified | 73.5% ↗epoch run | 48.3% ↗epoch run | +25.2 pt | Gemini 3.1 Pro Preview |
| tau2-bench Banking Knowledge (Sierra) | 26% ↗ | 47.9% ↗ | -21.9 pt | Grok 4.5 |
| Terminal-Bench 4.0 (Vals AI) | 2.5% ↗ | 8.6% ↗ | -6.1 pt | Grok 4.5 |
| Vending-Bench 2 (Andon Labs) | 3774.25 ↗ | 3887.43 ↗ | -113.2 | Grok 4.5 |
| WeirdML (Håvard Tveit Ihle) | 72.1% ↗ | 46.4% ↗ | +25.7 pt | Gemini 3.1 Pro Preview |
| Only Gemini 3.1 Pro Preview reports · 10 | ||||
| AIME 2026 | 98.3% ↗matharena ⚠ | not reported | — | — |
| ARC-AGI-2 | 77.1% ↗ | not reported | — | — |
| BALROG (BALROG) | 57% ↗ | not reported | — | — |
| EBR-bench (Epoch AI run) | 14.3% ↗epoch run | not reported | — | — |
| GSO Opt@1 (GSO) | 21.6% ↗ | not reported | — | — |
| MirrorCode (Epoch AI run) | 8.9% ↗epoch run | not reported | — | — |
| Mystery Game Puzzles (Epoch AI run) | 34% ↗epoch run | not reported | — | — |
| SWE-Bench Pro | 54.2% ↗ | not reported | — | — |
| SWE-bench Verified (Epoch AI run) | 75.6% ↗epoch run | not reported | — | — |
| Toolathlon-Verified (HKUST) | 61.1% ↗ | not reported | — | — |
| Only Grok 4.5 reports · 1 | ||||
| SWE-rebench 2026-05-15 to 2026-07-01 (Nebius) | not reported | 63.8% ↗ | — | — |
Scores tagged "epoch" or "matharena" are independent runs, used only where the lab has not published its own; ⚠ marks rows MathArena flags as released after the competition. Higher is better on every row. Δ is Gemini 3.1 Pro Preview minus Grok 4.5 in the benchmark's own unit. "Not reported" means the lab has not published that figure; it is not a zero. ↗ opens the source.
Specs & pricing
Side by side
Official pricing pages and model cards| Spec | Gemini 3.1 Pro Preview | Grok 4.5 | Edge |
|---|---|---|---|
| Context window | 1.0M tokens | 500K tokens | Gemini 3.1 Pro Preview |
| Max output | 66K tokens | — | — |
| Input price / 1M | $2 ↗ | $2 ↗ | Tie |
| Output price / 1M | $12 ↗ | $6 ↗ | Grok 4.5 |
| Cached input / 1M | $0.2 ↗ | $0.3 ↗ | Gemini 3.1 Pro Preview |
| Input + output / 1M Lower is cheaper. List prices; batch, tool and regional fees excluded. | $14 | $8 | Grok 4.5 |
| Modalities | text · vision · audio | text · vision | Gemini 3.1 Pro Preview |
| Released | 2026-02-19 | 2026-07-08 | — |
| Cited benchmark scores | 43 | 33 | — |
Reliability
Provider status
Live from /statusMore matchups
Gemini 3.1 Pro Preview vs …
Models sharing the most benchmarksMore matchups
Grok 4.5 vs …
Models sharing the most benchmarksBuilt by Respan
Which one wins on your data?
Public benchmarks are a starting point. Run Gemini 3.1 Pro Preview and Grok 4.5 on your own prompts with Respan evals, or route to either through one gateway key with automatic failover.