Compare · ModelsLive · 2 picked · head to head
GPT-5.1 vs GPT-5.4
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
GPT-5.4 wins on 19/22 benchmarks
GPT-5.4 wins 19 of 22 shared benchmarks. Leads in agentic · reasoning · arena.
Category leads
agentic·GPT-5.4reasoning·GPT-5.4arena·GPT-5.4knowledge·GPT-5.4general·GPT-5.4math·GPT-5.4coding·GPT-5.4
Head to head · from the data · data as of 2026-11-22
- Benchmarks · GPT-5.1 leads on 3 of 22 shared benchmarks (DeepResearch Bench, SimpleQA Verified, WeirdML), GPT-5.4 on 19 (APEX-Agents, ARC-AGI, ARC-AGI-2, Chatbot Arena Elo · Coding, 15 more)
- Price · GPT-5.1 input $1.25/M vs GPT-5.4 $2.50/M, GPT-5.1 is 50% cheaper (OpenRouter list price, as of 2026-11-22)
- Cheapest provider · GPT-5.1 at OpenAI $0.63/M input (of 4 offers) as of 2026-11-22 · GPT-5.4 at OpenAI $1.25/M input (of 4 offers) as of 2026-11-22
- Context · GPT-5.1 400K tokens vs GPT-5.4 1.1M tokens
- Released · GPT-5.1 Nov 13, 2025 vs GPT-5.4 Mar 5, 2026 · 4 months apart
- Price history · GPT-5.1 OpenRouter input list price had no move of 5% or more in the 90 days to 2026-11-22 · GPT-5.4 OpenRouter input list price had no move of 5% or more in the 90 days to 2026-11-22
Sources · scores from public leaderboards, list prices and provider offers from OpenRouter, price moves from the BenchGecko changelog (5% or more, held in the next daily check), Gecko tests run by BenchGecko.
Hype vs Reality
Attention vs performance
GPT-5.1
#159 by perf·#7 by attention
GPT-5.4
#112 by perf·#7 by attention
Best value
GPT-5.1
1.4x better value than GPT-5.4
GPT-5.1
8.4 pts/$
$5.63/M
GPT-5.4
6.0 pts/$
$8.75/M
Vendor risk
Who is behind the model
OpenAI
$840.0B·Tier 1
OpenAI
$840.0B·Tier 1
Head to head
22 benchmarks · 2 models
GPT-5.1GPT-5.4
APEX-Agents
GPT-5.4 leads by +34.9
APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments.
GPT-5.1
17.5
GPT-5.4
52.4
ARC-AGI
GPT-5.4 leads by +20.9
ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization.
GPT-5.1
72.8
GPT-5.4
93.7
ARC-AGI-2
GPT-5.4 leads by +56.3
ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data.
GPT-5.1
17.6
GPT-5.4
74.0
Chatbot Arena Elo · Coding
GPT-5.4 leads by +54.0
GPT-5.1
1341.0
GPT-5.4
1395.0
Chatbot Arena Elo · Overall
GPT-5.4 leads by +26.3
GPT-5.1
1438.5
GPT-5.4
1464.8
Chess Puzzles
GPT-5.4 leads by +6.3
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
GPT-5.1
39.0
GPT-5.4
45.3
Cl Bench
GPT-5.4 leads by +4.2
GPT-5.1
23.7
GPT-5.4
27.9
Cl Bench Life
GPT-5.4 leads by +4.4
GPT-5.1
17.3
GPT-5.4
21.7
DeepResearch Bench
GPT-5.1 leads by +7.7
DeepResearch Bench · evaluates AI on complex multi-step research tasks requiring information gathering, synthesis, and producing comprehensive analyses.
GPT-5.1
42.8
GPT-5.4
35.1
Dtbench
GPT-5.4 leads by +7.1
GPT-5.1
83.5
GPT-5.4
90.7
FrontierMath-2025-02-28-Private
GPT-5.4 leads by +16.6
FrontierMath (Feb 2025) · original research-level math problems created by mathematicians, testing capabilities at the boundary of current AI mathematical reasoning.
GPT-5.1
31.0
GPT-5.4
47.6
FrontierMath-Tier-4-2025-07-01-Private
GPT-5.4 leads by +14.6
FrontierMath Tier 4 (Jul 2025) · the most challenging tier of frontier mathematics, containing problems that push the absolute limits of AI mathematical reasoning.
GPT-5.1
12.5
GPT-5.4
27.1
GPQA diamond
GPT-5.4 leads by +7.9
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
GPT-5.1
87.0
GPT-5.4
94.9
GSO-Bench
GPT-5.4 leads by +17.6
GSO-Bench · evaluates AI models on real-world open-source software engineering tasks, testing the ability to understand and resolve actual GitHub issues.
GPT-5.1
13.7
GPT-5.4
31.4
HLE
GPT-5.4 leads by +13.2
HLE (Humanity's Last Exam) · a reasoning benchmark designed to be the hardest public evaluation of AI. Questions span mathematics, physics, philosophy, and logic · curated to be at or beyond the frontier of human expert capability. Tested with and without tool augmentation.
GPT-5.1
19.8
GPT-5.4
33.0
Lmca
GPT-5.4 leads by +9.5
GPT-5.1
51.6
GPT-5.4
61.1
Mystery Game Puzzles
GPT-5.4 leads by +19.8
GPT-5.1
10.8
GPT-5.4
30.6
OTIS Mock AIME 2024-2025
GPT-5.4 leads by +9.2
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
GPT-5.1
88.6
GPT-5.4
97.8
SimpleQA Verified
GPT-5.1 leads by +2.9
SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information.
GPT-5.1
48.0
GPT-5.4
45.1
SWE-Bench verified
GPT-5.4 leads by +8.9
SWE-bench Verified · 500 human-validated tasks from 12 real Python repositories (Django, Flask, scikit-learn, sympy, and others). Each task requires the model to produce a git patch that resolves a real GitHub issue and passes the test suite. The verified subset eliminates ambiguous tasks from the original SWE-bench. The benchmark remains the most-cited evaluation for code-generation capability.
GPT-5.1
68.0
GPT-5.4
76.9
Terminal Bench
GPT-5.4 leads by +34.2
Terminal-Bench 2.0 · evaluates AI agents on real terminal-based coding tasks · writing scripts, debugging, running tests, and managing projects entirely through command-line interaction. Tests both code quality and terminal fluency.
GPT-5.1
47.6
GPT-5.4
81.8
WeirdML
GPT-5.1 leads by +3.3
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
GPT-5.1
60.8
GPT-5.4
57.4
Full benchmark table
| Benchmark | GPT-5.1 | GPT-5.4 |
|---|---|---|
APEX-Agents APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments. | 17.5 | 52.4 |
ARC-AGI ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization. | 72.8 | 93.7 |
ARC-AGI-2 ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data. | 17.6 | 74.0 |
Chatbot Arena Elo · Coding | 1341.0 | 1395.0 |
Chatbot Arena Elo · Overall | 1438.5 | 1464.8 |
Chess Puzzles Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities. | 39.0 | 45.3 |
Cl Bench | 23.7 | 27.9 |
Cl Bench Life | 17.3 | 21.7 |
DeepResearch Bench DeepResearch Bench · evaluates AI on complex multi-step research tasks requiring information gathering, synthesis, and producing comprehensive analyses. | 42.8 | 35.1 |
Dtbench | 83.5 | 90.7 |
FrontierMath-2025-02-28-Private FrontierMath (Feb 2025) · original research-level math problems created by mathematicians, testing capabilities at the boundary of current AI mathematical reasoning. | 31.0 | 47.6 |
FrontierMath-Tier-4-2025-07-01-Private FrontierMath Tier 4 (Jul 2025) · the most challenging tier of frontier mathematics, containing problems that push the absolute limits of AI mathematical reasoning. | 12.5 | 27.1 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 87.0 | 94.9 |
GSO-Bench GSO-Bench · evaluates AI models on real-world open-source software engineering tasks, testing the ability to understand and resolve actual GitHub issues. | 13.7 | 31.4 |
HLE HLE (Humanity's Last Exam) · a reasoning benchmark designed to be the hardest public evaluation of AI. Questions span mathematics, physics, philosophy, and logic · curated to be at or beyond the frontier of human expert capability. Tested with and without tool augmentation. | 19.8 | 33.0 |
Lmca | 51.6 | 61.1 |
Mystery Game Puzzles | 10.8 | 30.6 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 88.6 | 97.8 |
SimpleQA Verified SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information. | 48.0 | 45.1 |
SWE-Bench verified SWE-bench Verified · 500 human-validated tasks from 12 real Python repositories (Django, Flask, scikit-learn, sympy, and others). Each task requires the model to produce a git patch that resolves a real GitHub issue and passes the test suite. The verified subset eliminates ambiguous tasks from the original SWE-bench. The benchmark remains the most-cited evaluation for code-generation capability. | 68.0 | 76.9 |
Terminal Bench Terminal-Bench 2.0 · evaluates AI agents on real terminal-based coding tasks · writing scripts, debugging, running tests, and managing projects entirely through command-line interaction. Tests both code quality and terminal fluency. | 47.6 | 81.8 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 60.8 | 57.4 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
GPT-5.1 vs GPT-5.4 · questions answered from the data
GPT-5.1 leads on 3 of 22 shared benchmarks (DeepResearch Bench, SimpleQA Verified, WeirdML), GPT-5.4 on 19 (APEX-Agents, ARC-AGI, ARC-AGI-2, Chatbot Arena Elo · Coding, 15 more)