# GPT-4 Turbo vs GPT-5.4

> GPT-5.4 is the stronger model overall, scoring 59.4 to 30.5 on the Noometry Index.

- Canonical page: https://noometry.com/compare/gpt-4-turbo-vs-gpt-5-4
- Last updated: 2026-10-10
- Shared benchmarks: 28

## Summary

- They share 28 benchmarks with published results for both. GPT-4 Turbo scores higher in 0 categories and GPT-5.4 in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5.4 leads 73.5 to 9.0.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 6.7% for GPT-4 Turbo and 97.8% for GPT-5.4.
- GPT-5.4 is cheaper at $2.50 / $15 per million input/output tokens, against $10 / $30 for GPT-4 Turbo.
- GPT-5.4 accepts more context: 1.05M tokens versus 128K.

## Snapshot

| | GPT-4 Turbo | GPT-5.4 |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 30.5 | 59.4 |
| Rank | 292 | 16 |
| Context | 128K | 1.05M |
| Input $/M | $10 | $2.50 |
| Output $/M | $30 | $15 |
| Weights | Proprietary | Proprietary |

## Coding

- GPT-4 Turbo: 33.8 (#249)
- GPT-5.4: 52.6 (#33)

| Benchmark | GPT-4 Turbo | GPT-5.4 |
|---|---|---|
| WeirdML | 18% | 77.7% |
| LMArena Coding | 1268 | 1497 |
| SWE-bench Verified | — | 76.9% |
| DeepSWE | — | 51.8% |
| LMArena WebDev | — | 1465 |
| SciCode | — | 56.6% |
| GSO | — | 31.4% |
| BigCodeBench Instruct | 48.2% | — |
| MirrorCode | — | 15.6% |
| BigCodeBench Complete | 58.2% | — |
| ALE-Bench | — | 1,607 |
| AlgoTune | — | 1.85 |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73.3% | — |

## Agentic & Tool Use

- GPT-4 Turbo: —
- GPT-5.4: 46.5 (#13)

| Benchmark | GPT-4 Turbo | GPT-5.4 |
|---|---|---|
| METR Time Horizons | 36.7% | 74.3% |
| Terminal-Bench | — | 81.8% |
| APEX-Agents | — | 52.4% |
| τ²-bench Banking | — | 39.4% |
| DeepResearch Bench | — | 35.1% |
| PostTrainBench | — | 19% |
| GBAEval | — | 45.1% |
| LMArena Search | — | 1197 |
| Vending-Bench 2 | — | 6,144 |

## Reasoning

- GPT-4 Turbo: 15.3 (#317)
- GPT-5.4: 61.8 (#19)

| Benchmark | GPT-4 Turbo | GPT-5.4 |
|---|---|---|
| Chess Puzzles | 6% | 44% |
| LMArena Hard Prompts | 1251 | 1485 |
| DTBench | 61.6% | 94.4% |
| LMCA | 9.8% | 52% |
| Epoch Capabilities Index | 127.25 | 156.81 |
| ForecastBench | 59.4 | 59.5 |
| ARC-AGI-2 | — | 74% |
| SimpleBench | 25.1% | — |
| Kagi LLM Benchmark | — | 63.8% |
| NYT Connections (extended) | — | 91.3% |
| ARC-AGI-1 | — | 93.7% |
| CritPt | — | 23.4% |
| EnigmaEval | — | 16% |
| Thematic Generalization | — | 80% |
| EBR-Bench | — | 25.4% |
| Mystery Game Puzzles | — | 37% |

## Math

- GPT-4 Turbo: 9.0 (#322)
- GPT-5.4: 73.5 (#19)

| Benchmark | GPT-4 Turbo | GPT-5.4 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 0.7% | 78.6% |
| OTIS Mock AIME 2024-2025 | 6.7% | 97.8% |
| LMArena Math | 1272 | 1488 |
| FrontierMath Tier 4 | — | 49% |
| MathArena Final-Answer Competitions | — | 83.1% |
| ProofBench | — | 56% |
| MATH Level 5 | 46.7% | — |
| FrontierMath (Feb 2025 set) | — | 47.6% |
| FrontierMath Tier 4 (v1) | — | 27.1% |

## Knowledge

- GPT-4 Turbo: 24.3 (#268)
- GPT-5.4: 65.3 (#14)

| Benchmark | GPT-4 Turbo | GPT-5.4 |
|---|---|---|
| GPQA Diamond | 46.6% | 93.3% |
| LMArena Expert | 1223 | 1507 |
| Humanity's Last Exam | — | 36.2% |
| SimpleQA Verified | — | 45.1% |
| Confabulations | 28.4% | — |
| Vectara Hallucination Rate | — | 7% |
| MMLU | 81.3% | — |

## Multimodal

- GPT-4 Turbo: 30.6 (#110)
- GPT-5.4: 43.7 (#20)

| Benchmark | GPT-4 Turbo | GPT-5.4 |
|---|---|---|
| LMArena Vision | 1090 | 1303 |
| Blueprint-Bench 2 | — | 27.1% |
| Furniture Assembly | — | 37.5% |
| LMArena Document | — | 1471 |

## Multilingual

- GPT-4 Turbo: 40.5 (#216)
- GPT-5.4: 56.2 (#23)

| Benchmark | GPT-4 Turbo | GPT-5.4 |
|---|---|---|
| LMArena Non-English | 1245 | 1465 |
| LMArena Chinese | 1242 | 1519 |
| LMArena French | 1276 | 1493 |
| LMArena German | 1259 | 1472 |
| LMArena Japanese | 1194 | 1485 |
| LMArena Korean | 1187 | 1448 |
| LMArena Russian | 1259 | 1480 |
| LMArena Spanish | 1260 | 1454 |

## Instruction Following

- GPT-4 Turbo: 65.8 (#216)
- GPT-5.4: 77.1 (#27)

| Benchmark | GPT-4 Turbo | GPT-5.4 |
|---|---|---|
| LMArena Instruction Following | 1249 | 1469 |

## Long Context

- GPT-4 Turbo: 38.0 (#206)
- GPT-5.4: 50.3 (#8)

| Benchmark | GPT-4 Turbo | GPT-5.4 |
|---|---|---|
| LMArena Longer Query | 1254 | 1473 |
| CL-bench | — | 27.9% |
| CL-bench Life | — | 21.7% |

## Writing & Preference

- GPT-4 Turbo: 47.7 (#206)
- GPT-5.4: 71.9 (#17)

| Benchmark | GPT-4 Turbo | GPT-5.4 |
|---|---|---|
| LMArena Text | 1272 | 1469 |
| LMArena Creative Writing | 1269 | 1439 |
| LMArena Multi-Turn | 1267 | 1482 |
| EQ-Bench Creative Writing | — | 1840 |
| EQ-Bench 4 | — | 1272 |

## FAQ

### Is GPT-4 Turbo better than GPT-5.4?

GPT-5.4 is the stronger model overall, scoring 59.4 to 30.5 on the Noometry Index.

### Which is cheaper, GPT-4 Turbo or GPT-5.4?

GPT-5.4 is cheaper. It lists at $2.50 per million input tokens and $15 per million output tokens; GPT-4 Turbo lists at $10 and $30.

### Is GPT-4 Turbo or GPT-5.4 better for coding?

GPT-5.4 scores higher on coding benchmarks: 52.6 versus 33.8 in the Noometry coding category.

### Which has the bigger context window?

GPT-5.4 does, with 1.05M tokens against 128K.

### How many benchmarks do GPT-4 Turbo and GPT-5.4 share?

28 benchmarks have published results for both models. GPT-4 Turbo has 36 scored results on Noometry and GPT-5.4 has 68.