GPT-5.4vsGPT-5.1
Across 10 shared benchmarks, GPT-5.4 leads overall: GPT-5.4 wins 9, GPT-5.1 wins 1, with 0 ties and an average score difference of +18.12.
GPT-5.49 wins(90%)(10%)1 winGPT-5.1
Benchmark scores
Grouped by capability, sorted by largest gap within each. 10 shared benchmarks.
General Knowledge
GPT-5.4 4/4| Benchmark | GPT-5.4 | GPT-5.1 | Diff |
|---|---|---|---|
| ARC-AGI-2 | 77.107 / 59Normal (No Tools) | 17.6033 / 59 | +59.50 |
| HLE | 52.1014 / 157极高强度思考（工具） | 26.5083 / 157 | +25.60 |
| ARC-AGI | 93.707 / 65Normal (No Tools) | 72.8025 / 65 | +20.90 |
| GPQA Diamond | 92.8010 / 178极高强度思考（无工具） | 88.1028 / 178 | +4.70 |
Math and Reasoning
GPT-5.4 2/2| Benchmark | GPT-5.4 | GPT-5.1 | Diff |
|---|---|---|---|
| FrontierMath | 47.605 / 60极高强度思考（无工具） | 26.7013 / 60Thinking High (With Tools) | +20.90 |
| FrontierMath - Tier 4 | 27.1011 / 80极高强度思考（无工具） | 12.5029 / 80Thinking High (With Tools) | +14.60 |
Agent Level Benchmark
GPT-5.1 1/1| Benchmark | GPT-5.4 | GPT-5.1 | Diff |
|---|---|---|---|
| τ²-Bench - Telecom | 64.3030 / 35Normal (With Tools) | 95.6014 / 35Thinking High (With Tools) | -31.30 |
AI Agent - Information Search
GPT-5.4 1/1| Benchmark | GPT-5.4 | GPT-5.1 | Diff |
|---|---|---|---|
| BrowseComp | 82.7011 / 45极高强度思考（工具） | 50.8036 / 45Thinking High (No Tools) | +31.90 |
AI Agent - Tool Usage
GPT-5.4 1/1| Benchmark | GPT-5.4 | GPT-5.1 | Diff |
|---|---|---|---|
| Terminal Bench 2.0 | 75.104 / 46极高强度思考（工具） | 47.6037 / 46Thinking High (With Tools) | +27.50 |
Coding and Software Engineer
GPT-5.4 1/1| Benchmark | GPT-5.4 | GPT-5.1 | Diff |
|---|---|---|---|
| SWE-Bench Pro - Public | 57.7010 / 43极高强度思考（无工具） | 50.8030 / 43Thinking High (No Tools) | +6.90 |
Specs
| Field | GPT-5.4 | GPT-5.1 |
|---|---|---|
| Publisher | OpenAI | OpenAI |
| Release date | 2026-03-05 | 2025-11-12 |
| Model type | Multimodal model | Reasoning model |
| Architecture | Dense | Dense |
| Parameters | Not available | Not available |
| Context length | 1M | 400K |
| Max output | 125K | 128K |
API pricing
Prices use DataLearner records when available; missing fields are not inferred.
| Item | GPT-5.4 | GPT-5.1 |
|---|---|---|
| Text input | $2.5 / 1M tokens | Not public |
| Text output | $15 / 1M tokens | Not public |
| Cache write | $0.25 / 1M tokens | Not public |
One or both models have incomplete public pricing.
Summary
- GPT-5.4leads in:General Knowledge (4/4), Math and Reasoning (2/2), AI Agent - Information Search (1/1), AI Agent - Tool Usage (1/1), Coding and Software Engineer (1/1)
- GPT-5.1leads in:Agent Level Benchmark (1/1)
On average across the 10 shared benchmarks, GPT-5.4 scores 18.12 higher.
Largest single-benchmark gap: ARC-AGI-2 — GPT-5.4 77.10 vs GPT-5.1 17.60 (+59.50).
Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.