Grouped by capability, sorted by largest gap within each. 13 shared benchmarks.
| Benchmark | GPT-5.1 | GPT-5 | Diff |
|---|---|---|---|
| ARC-AGI | 72.8025 / 65high | 661 / 65 | +66.80 |
| HLE | 42.7038 / 149Thinking High (With Tools + Internet) | 6.30138 / 149 | +36.40 |
| ARC-AGI-2 | 17.6032 / 58high | 056 / 58 | +17.60 |
| GPQA Diamond | 88.1025 / 175Thinking High (No Tools) | 77.8081 / 175 | +10.30 |
| Benchmark | GPT-5.1 | GPT-5 | Diff |
|---|---|---|---|
| AIME2025 | 9428 / 106Thinking High (No Tools) | 61.9080 / 106 | +32.10 |
| Simple Bench | 53.2010 / 27high | 56.708 / 27high | -3.50 |
| FrontierMath | 26.7013 / 60Thinking High (With Tools) | 26.3014 / 60Thinking High (With Tools) | +0.40 |
| FrontierMath - Tier 4 | 12.5029 / 80 |
| Benchmark | GPT-5.1 | GPT-5 | Diff |
|---|---|---|---|
| SWE-Bench Pro - Public | 50.8024 / 36Thinking High (No Tools) | 36.3034 / 36high | +14.50 |
| SWE-bench Verified | 76.3025 / 103high | 72.8041 / 103high | +3.50 |
| Benchmark | GPT-5.1 | GPT-5 | Diff |
|---|---|---|---|
| τ²-Bench - Telecom | 95.6014 / 35Thinking High (With Tools) | 96.7011 / 35Thinking High (With Tools) | -1.10 |
| Benchmark | GPT-5.1 | GPT-5 | Diff |
|---|---|---|---|
| BrowseComp | 50.8034 / 43Thinking High (No Tools) | 54.9030 / 43thinking + 使用工具 | -4.10 |
| Benchmark | GPT-5.1 | GPT-5 | Diff |
|---|---|---|---|
| MMMU | 85.402 / 28Thinking High (No Tools) | 84.205 / 28high | +1.20 |
| Field | GPT-5.1 | GPT-5 |
|---|---|---|
| Publisher | OpenAI | OpenAI |
| Release date | 2025-11-12 | 2025-08-07 |
| Model type | Reasoning model | Foundation model |
| Architecture | Dense | Dense |
| Parameters | 0.0 | 0.0 |
| Context length | 400K | 400K |
| Max output | 131072 | 131072 |
Prices use DataLearner records when available; missing fields are not inferred.
| Item | GPT-5.1 | GPT-5 |
|---|---|---|
| Text input | 1.25 美元/100万 tokens | 1.25 美元/100 万tokens |
| Text output | 10 美元/100万 tokens | 10 美元/100 万tokens |
| Cache read | 0.125 美元/100万 tokens | Not public |
On average across the 13 shared benchmarks, GPT-5.1 scores 13.39 higher.
Largest single-benchmark gap: ARC-AGI — GPT-5.1 72.80 vs GPT-5 6 (+66.80).
Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.
| 12.5029 / 80Thinking High (No Tools) |
| — |
