Compare / head-to-head
Claude Sonnet 4.5vsGemini 3.1 Pro Preview
Gemini 3.1 Pro Preview leads 5 of 5 shared benchmarks. Gemini 3.1 Pro Preview is 1.3x 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
0 – 5
Gemini 3.1 Pro Preview leads
Cheaper per token
Gemini 3.1 Pro Preview
1.3x cheaper, input + output
Larger context
Gemini 3.1 Pro Preview
1.0M tokens
Provider uptime (30d)
100% · 100%
Anthropic · Google
Claude Sonnet 4.5
Anthropic · released 2025-09-29
textvision
Gemini 3.1 Pro Preview
Google · released 2026-02-19
textvisionaudio
Quality
Benchmark matrix
5 shared · 15 only Claude Sonnet 4.5 · 4 only Gemini 3.1 Pro Preview| Benchmark | Claude Sonnet 4.5 | Gemini 3.1 Pro Preview | Δ | Edge |
|---|---|---|---|---|
| Reported by both · 5 | ||||
| GPQA Diamond | 83.4% ↗ | 94.3% ↗ | -10.9 pt | Gemini 3.1 Pro Preview |
| FrontierMath Tier 4 v2 (Epoch AI run) | 2.4% ↗epoch | 26.8% ↗epoch | -24.4 pt | Gemini 3.1 Pro Preview |
| FrontierMath Tiers 1-3 v2 (Epoch AI run) | 23.9% ↗epoch | 59.6% ↗epoch | -35.7 pt | Gemini 3.1 Pro Preview |
| OTIS Mock AIME 2024-2025 (Epoch AI run) | 77.8% ↗epoch | 95.6% ↗epoch | -17.8 pt | Gemini 3.1 Pro Preview |
| SimpleQA Verified | 30.7% ↗epoch | 73.5% ↗epoch | -42.8 pt | Gemini 3.1 Pro Preview |
| Only Claude Sonnet 4.5 reports · 15 | ||||
| SWE-bench Verified | 77.2% ↗ | not reported | — | — |
| AIME 2025 | 84.2% ↗matharena ⚠ | not reported | — | — |
| Humanity's Last Exam (no tools) | 17.7% ↗ | not reported | — | — |
| ARC-AGI-2 (Verified) | 13.6% ↗ | not reported | — | — |
| GDPval-AA | 1276 ↗ | not reported | — | — |
| Humanity's Last Exam (with tools) | 33.6% ↗ | not reported | — | — |
| MCP Atlas | 43.8% ↗ | not reported | — | — |
| MMMLU | 89.5% ↗ | not reported | — | — |
| MMMU-Pro (no tools) | 63.4% ↗ | not reported | — | — |
| MMMU-Pro (with tools) | 68.9% ↗ | not reported | — | — |
| OSWorld-Verified | 61.4% ↗ | not reported | — | — |
| SWE-bench Verified (Epoch AI run) | 71.3% ↗epoch | not reported | — | — |
| Tau2-bench Retail | 86.2% ↗ | not reported | — | — |
| Tau2-bench Telecom | 98% ↗ | not reported | — | — |
| Terminal-Bench 2.0 | 51% ↗ | not reported | — | — |
| Only Gemini 3.1 Pro Preview reports · 4 | ||||
| AIME 2026 | not reported | 98.3% ↗matharena ⚠ | — | — |
| LMArena Elo | not reported | 1480.1 ↗ | — | — |
| ARC-AGI-2 | not reported | 77.1% ↗ | — | — |
| SWE-Bench Pro | not reported | 54.2% ↗ | — | — |
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 Claude Sonnet 4.5 minus Gemini 3.1 Pro Preview 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 | Claude Sonnet 4.5 | Gemini 3.1 Pro Preview | Edge |
|---|---|---|---|
| Context window | 200K tokens | 1.0M tokens | Gemini 3.1 Pro Preview |
| Max output | 64K tokens | 66K tokens | Gemini 3.1 Pro Preview |
| Input price / 1M | $3 ↗ | $2 ↗ | Gemini 3.1 Pro Preview |
| Output price / 1M | $15 ↗ | $12 ↗ | Gemini 3.1 Pro Preview |
| Cached input / 1M | $0.3 ↗ | $0.2 ↗ | Gemini 3.1 Pro Preview |
| Input + output / 1M Lower is cheaper. List prices; batch, tool and regional fees excluded. | $18 | $14 | Gemini 3.1 Pro Preview |
| Modalities | text · vision | text · vision · audio | Gemini 3.1 Pro Preview |
| Released | 2025-09-29 | 2026-02-19 | — |
| Cited benchmark scores | 21 | 10 | — |
Reliability
Provider status
Last 60 days · refreshed daily on this page · live on /statusMore matchups
Claude Sonnet 4.5 vs …
Models sharing the most benchmarksMore matchups
Gemini 3.1 Pro Preview vs …
Models sharing the most benchmarksBuilt by Respan
Which one wins on your data?
Public benchmarks are a starting point. Run Claude Sonnet 4.5 and Gemini 3.1 Pro Preview on your own prompts with Respan evals, or route to either through one gateway key with automatic failover.