GPT-5.4
| Released | 2026-03-05 |
|---|---|
| Price per 1M | $2.50 in · $15.00 out |
| Overall score | 58.9 · rank 39 |
| Context window | 1,050,000 tokens |
| Provider | OpenAI |
| Weights | Proprietary |
Scores
| Dimension | Score | Rank | Benchmarks |
|---|---|---|---|
| Overall: Balanced | 58.9 | 39 | 42 |
| Overall: Coding | 57.8 | 52 | 26 |
| Overall: Reasoning | 60.1 | 33 | 18 |
| Overall: Agentic work | 58.6 | 46 | 24 |
| Overall: Frontend | 56.5 | 51 | 11 |
| Coding | 57.7 | 49 | 7 |
| Agents | 58.3 | 50 | 7 |
| Reasoning | 61.7 | 30 | 10 |
| Math | 60.1 | 30 | 5 |
| Knowledge | 56.3 | 51 | 3 |
| Frontend | 55.2 | 53 | 2 |
| General | 61.2 | 23 | 6 |
| Multimodal | 60.2 | 18 | 2 |
| Value vs price | +5.9 | 65 |
Configurations
| Configuration | Balanced | Benchmarks |
|---|---|---|
| Unspecified setting | 56.3 ±3.6 | 12 |
| High | 57.9 ±2.2 | 15 |
| Low | 47.9 ±4.5 | 7 |
| Medium | 58.4 ±4.2 | 9 |
| Non-reasoning | 49.0 ±4.7 | 6 |
| Extra High | 59.1 ±1.0 | 29 |
Evidence
Coding 57.7 ±1.3
| Benchmark | Result | Source | Date |
|---|---|---|---|
| DeepSWE | 51.8% | Epoch AI | 2026-11-11 |
| MirrorCode | 23.3% | Epoch AI | 2026-10-29 |
| GSO | 31.4% | Epoch AI | 2026-11-11 |
| SciCode | 56.6% | Epoch AI | 2026-11-11 |
| LiveCodeBench (Vals run) | 84.1% | Vals AI | 2026-09-01 |
| SWE-bench Verified (Epoch run) | 76.9% | Epoch AI | 2026-03-06 |
| ProgramBench | 0.5% | Vals AI | 2026-11-09 |
| SWE-bench Verified (Vals run) | 78.2% | Vals AI | 2026-09-01 |
Agents 58.3 ±3.3
| Benchmark | Result | Source | Date |
|---|---|---|---|
| APEX-Agents | 52.4% | Epoch AI | 2026-11-11 |
| Terminal-Bench 2.0 | 81.8% | Epoch AI | 2026-03-12 |
| METR 50% Time Horizon | $342 | METR | 2026-11-11 |
| tau2-bench Banking Knowledge | 39.4% | tau2-bench | 2026-05-06 |
| METR task suite | 74.3% | Epoch AI | 2026-11-11 |
| SkillsBench | 51.7% | Vals AI | 2026-10-29 |
| DeepResearch Bench | 35.1% | Epoch AI | 2026-11-11 |
| Vending-Bench 2 | $6,144 | Epoch AI | 2026-11-11 |
Reasoning 61.7 ±1.8
| Benchmark | Result | Source | Date |
|---|---|---|---|
| ARC-AGI-2 | 74.0% | Epoch AI | 2026-11-11 |
| ARC-AGI-2 (ARC Prize verified) | 74.0% | ARC Prize | 2026-11-11 |
| Extended NYT Connections | 91.3% | Lech Mazur | 2026-11-10 |
| CritPt | 23.4% | Epoch AI | 2026-11-11 |
| ARC-AGI-1 (ARC Prize verified) | 93.7% | ARC Prize | 2026-11-11 |
| ARC-AGI-1 | 93.7% | Epoch AI | 2026-11-11 |
| EnigmaEval | 16.0% | Epoch AI | 2026-11-11 |
| GPQA Diamond | 93.3% | Epoch AI | 2026-03-06 |
| GPQA Diamond (Vals run) | 91.7% | Vals AI | 2026-09-01 |
| ARC-AGI-3 | 0.2% | ARC Prize | 2026-11-11 |
| Kagi Reasoning Benchmark | 63.8% | Kagi | 2026-11-11 |
| Chess Puzzles | 48.0% | Epoch AI | 2026-10-13 |
| Mystery Game Puzzles | 37.0% | Epoch AI | 2026-07-24 |
Math 60.1 ±1.2
| Benchmark | Result | Source | Date |
|---|---|---|---|
| FrontierMath Tiers 1-3 | 78.6% | Epoch AI | 2026-06-11 |
| FrontierMath Tier 4 | 49.0% | Epoch AI | 2026-06-11 |
| ProofBench | 56.0% | Epoch AI | 2026-11-11 |
| OTIS Mock AIME 2024-2025 | 97.8% | Epoch AI | 2026-07-15 |
| MathArena Final-Answer Competitions | 83.1% | MathArena | 2026-11-11 |
| FrontierMath Tiers 1-3 (2025 set) | 47.6% | Epoch AI | 2026-03-06 |
| FrontierMath Tier 4 (2025 set) | 27.1% | Epoch AI | 2026-03-06 |
Knowledge 56.3 ±2.6
| Benchmark | Result | Source | Date |
|---|---|---|---|
| SimpleQA Verified | 45.1% | Epoch AI | 2026-08-27 |
| Humanity's Last Exam | 36.2% | Epoch AI | 2026-11-11 |
| MMLU-Pro (Vals run) | 87.5% | Vals AI | 2026-09-01 |
Frontend 55.2 ±0.7
| Benchmark | Result | Source | Date |
|---|---|---|---|
| Vibe Code Bench v1.1 | 67.4% | Vals AI | 2026-11-09 |
| LMArena WebDev | 1,462 | LMArena | 2026-11-09 |
General 61.2 ±1.6
| Benchmark | Result | Source | Date |
|---|---|---|---|
| LLM Debate Benchmark | 1,540 | Lech Mazur | 2026-10-26 |
| EQ-Bench 4 | 1,258 | EQ-Bench | 2026-10-24 |
| EQ-Bench Creative Writing v3 | 1,841 | EQ-Bench | 2026-11-11 |
| LMArena Text | 1,475 | LMArena | 2026-11-09 |
| Creative Story-Writing (Lech Mazur) | 77.0% | Lech Mazur | 2026-10-23 |
| EQ-Bench Longform Creative Writing | 78.3% | EQ-Bench | 2026-11-11 |
Multimodal 60.2 ±2.3
| Benchmark | Result | Source | Date |
|---|---|---|---|
| MMMU-Pro (Vals run) | 87.5% | Vals AI | 2026-09-01 |
| LMArena Vision | 1,282 | LMArena | 2026-11-09 |
Results not used in the score
| Benchmark | Result | Source | Why not used |
|---|---|---|---|
| ARC-AGI-1 | 86.2% | Epoch AI | A stronger configuration result was used |
| ARC-AGI-1 | 92.7% | Epoch AI | A stronger configuration result was used |
| ARC-AGI-1 | 68.2% | Epoch AI | A stronger configuration result was used |
| ARC-AGI-1 (ARC Prize verified) | 86.2% | ARC Prize | A stronger configuration result was used |
| ARC-AGI-1 (ARC Prize verified) | 92.7% | ARC Prize | A stronger configuration result was used |
| ARC-AGI-1 (ARC Prize verified) | 68.2% | ARC Prize | A stronger configuration result was used |
| ARC-AGI-2 | 55.4% | Epoch AI | A stronger configuration result was used |
| ARC-AGI-2 | 67.5% | Epoch AI | A stronger configuration result was used |
| ARC-AGI-2 | 29.2% | Epoch AI | A stronger configuration result was used |
| ARC-AGI-2 (ARC Prize verified) | 55.4% | ARC Prize | A stronger configuration result was used |
| ARC-AGI-2 (ARC Prize verified) | 67.5% | ARC Prize | A stronger configuration result was used |
| ARC-AGI-2 (ARC Prize verified) | 29.2% | ARC Prize | A stronger configuration result was used |
| Chess Puzzles | 37.0% | Epoch AI | A stronger configuration result was used |
| Chess Puzzles | 3.0% | Epoch AI | A stronger configuration result was used |
| Chess Puzzles | 42.0% | Epoch AI | A stronger configuration result was used |
| Chess Puzzles | 19.0% | Epoch AI | A stronger configuration result was used |
| Creative Story-Writing (Lech Mazur) | 76.0% | Lech Mazur | A stronger configuration result was used |
| Extended NYT Connections | 87.8% | Lech Mazur | A stronger configuration result was used |
| Extended NYT Connections | 17.8% | Lech Mazur | A stronger configuration result was used |
| Extended NYT Connections | 90.6% | Lech Mazur | A stronger configuration result was used |
| GPQA Diamond | 88.9% | Epoch AI | A stronger configuration result was used |
| GPQA Diamond | 74.8% | Epoch AI | A stronger configuration result was used |
| GPQA Diamond | 89.9% | Epoch AI | A stronger configuration result was used |
| GPQA Diamond | 84.9% | Epoch AI | A stronger configuration result was used |
| GSO | 25.5% | Epoch AI | A stronger configuration result was used |
| LLM Debate Benchmark | 1,487 | Lech Mazur | A stronger configuration result was used |
| LMArena Text | 1,465 | LMArena | A stronger configuration result was used |
| LMArena Vision | 1,278 | LMArena | A stronger configuration result was used |
| LMArena WebDev | 1,399 | LMArena | A stronger configuration result was used |
| LMArena WebDev | 1,442 | LMArena | A stronger configuration result was used |
| MirrorCode | 15.6% | Epoch AI | A stronger configuration result was used |
| Mystery Game Puzzles | 28.0% | Epoch AI | A stronger configuration result was used |
| Mystery Game Puzzles | 16.0% | Epoch AI | A stronger configuration result was used |
| Mystery Game Puzzles | 17.0% | Epoch AI | A stronger configuration result was used |
| OTIS Mock AIME 2024-2025 | 95.6% | Epoch AI | A stronger configuration result was used |
| OTIS Mock AIME 2024-2025 | 57.8% | Epoch AI | A stronger configuration result was used |
| OTIS Mock AIME 2024-2025 | 84.4% | Epoch AI | A stronger configuration result was used |
| OTIS Mock AIME 2024-2025 | 95.3% | Epoch AI | A stronger configuration result was used |
| ProgramBench | 0.0% | Vals AI | A stronger configuration result was used |
| Vibe Code Bench v1.1 | 48.5% | Vals AI | A stronger configuration result was used |
Names matched to this model
gpt-5-4, gpt-5-4-2026-03-05