# Claude Opus 5.5 vs GPT-5.4

> Side-by-side benchmark comparison for Claude Opus 5.5 and GPT-5.4 across agentic, coding, multimodal, knowledge, reasoning, multilingual, and math tasks.

- Canonical page: https://benchlm.ai/compare/claude-opus-5-5-vs-gpt-5-4
- Last updated: September 22, 2026

- Shared sourced benchmarks: 4
- HTML indexing: indexable
- Ranking lane: BenchAlign v5

## Quick Verdict

Pick Claude Opus 5.5 if you want the stronger benchmark profile. GPT-5.4 only makes more sense when its price, context window, or workload-specific category wins matter more than the overall score.

## Summary

- Claude Opus 5.5 leads overall 81.03 to 71.04.
- The clearest category separation is in agentic, where the averages are 82.9 for Claude Opus 5.5 and 52 for GPT-5.4.
- The biggest single benchmark swing is SWE-bench Pro in Coding, with scores of 89.9% and 57.7%.
- GPT-5.4 is the cheaper option on output tokens, which matters if you expect large responses or heavy interactive use.
- GPT-5.4 also has the larger context window at 1.05M.

## Model Snapshot

| Property | Claude Opus 5.5 | GPT-5.4 |
|----------|----------|----------|
| Creator | Anthropic | OpenAI |
| Type | Proprietary | Proprietary |
| Reasoning | Reasoning | Reasoning |
| Context | 1M | 1.05M |
| Overall Score | 81.03 | 71.04 |
| Benchmarks Covered | 47 | 38 |
| Pricing (input/output) | $4.00 / $20.00 | $2.50 / $15.00 |

## Category Breakdown

### Agentic

- Winner: Claude Opus 5.5
- Claude Opus 5.5 public-lane score: 82.9 (Supported · #1/157)
- GPT-5.4 public-lane score: 52 (Supported · #48/157)

| Benchmark | Claude Opus 5.5 | GPT-5.4 | Winner |
|-----------|-----------|-----------|--------|
| Terminal-Bench 4.0 | 66.40% | Coming soon | Coming soon |
| Terminal-Bench-Science 0.1 | 58.7% | Coming soon | Coming soon |
| AutomationBench | 40.0% | Coming soon | Coming soon |
| HLE w/ tools | 67.7% | Coming soon | Coming soon |
| OSWorld 2.0 | 48.7% | Coming soon | Coming soon |
| LAB all-pass (Harvey held-out) | 8.3% | Coming soon | Coming soon |
| LAB criterion-pass (Harvey held-out) | 91.2% | Coming soon | Coming soon |
| Toolathlon-Verified | 77.8% | Coming soon | Coming soon |
| Toolathlon Verified Pass@3 | 82.4% | Coming soon | Coming soon |
| Toolathlon Verified Pass³ | 72.2% | Coming soon | Coming soon |
| Toolathlon Verified avg. turns | 26.9 turns | Coming soon | Coming soon |
| Terminal-Bench 2.0 | Coming soon | 75.1% | Coming soon |
| CyberGym | Coming soon | 79.0% | Coming soon |
| BrowseComp | Coming soon | 82.7% | Coming soon |
| OSWorld-Verified | Coming soon | 75% | Coming soon |
| MCP Atlas | Coming soon | 70.6% | Coming soon |
| Toolathlon | Coming soon | 54.6% | Coming soon |
| τ²-bench results | Coming soon | 98.9% | Coming soon |
| Claw-Eval | Coming soon | 60.3% | Coming soon |
| DeepSearchQA | Coming soon | 73.6% | Coming soon |
| Gert Labs | Coming soon | 64.89% | Coming soon |
| ResearchClawBench | Coming soon | 15.3% | Coming soon |
| JobBench | Coming soon | 38.9% | Coming soon |
| ExploitGym | Coming soon | 6.0% | Coming soon |
| ApprenticeBench | Coming soon | 11% | Coming soon |

### Coding

- Winner: Claude Opus 5.5
- Claude Opus 5.5 public-lane score: 80.1 (Supported · #2/159)
- GPT-5.4 public-lane score: 54 (Supported · #45/159)

| Benchmark | Claude Opus 5.5 | GPT-5.4 | Winner |
|-----------|-----------|-----------|--------|
| FrontierCode 1.1 Main | 54.4% | Coming soon | Coming soon |
| cursorBench40 | 57.8% | Coming soon | Coming soon |
| SWE-bench Pro | 89.9% | 57.7% | Claude Opus 5.5 |
| SWE Multilingual | 93.9% | Coming soon | Coming soon |
| SWE Multimodal | 61.4% | Coming soon | Coming soon |
| DeepSWE | 74.2% | Coming soon | Coming soon |
| FrontierCode 1.1 Extended | 63.6% | Coming soon | Coming soon |
| FrontierSWE v2 | 62.3% | Coming soon | Coming soon |
| ProgramBench | 91.2% | Coming soon | Coming soon |
| LiveCodeBench Pro | Coming soon | 87.5% | Coming soon |
| React Native Evals | Coming soon | 85.3% | Coming soon |
| Vibe Code Bench | Coming soon | 67.42% | Coming soon |

### Reasoning

- Winner: Coming soon
- Claude Opus 5.5 public-lane score: 78.5 (#4/17)
- GPT-5.4 public-lane score: 57.2 (#14/17)

| Benchmark | Claude Opus 5.5 | GPT-5.4 | Winner |
|-----------|-----------|-----------|--------|
| ARC-AGI-2 | Coming soon | 74.0% | Coming soon |
| ARC-AGI-3 | Coming soon | 0.2% | Coming soon |

### Multimodal & Grounded

- Winner: Coming soon
- Claude Opus 5.5 public-lane score: 88.8 (#3/50)
- GPT-5.4 public-lane score: 69.3 (#21/50)

| Benchmark | Claude Opus 5.5 | GPT-5.4 | Winner |
|-----------|-----------|-----------|--------|
| Chartography (tools) | 89.0% | Coming soon | Coming soon |
| Chartography (no tools) | 64.4% | Coming soon | Coming soon |
| BenchCAD Vision2Code (no tools) | 0.730 | Coming soon | Coming soon |
| BenchCAD Vision2Code (tools) | 0.962 | Coming soon | Coming soon |
| Biomedical image analysis | 71.4% | Coming soon | Coming soon |
| OfficeQA | 78.9% | Coming soon | Coming soon |
| OfficeQA Pro | 67.7% | 53.2% | Claude Opus 5.5 |
| MMMU-Pro | Coming soon | 81.2% | Coming soon |
| MMMU-Pro w/ Python | Coming soon | 82.1% | Coming soon |
| CharXiv | Coming soon | 82.8% | Coming soon |
| ERQA | Coming soon | 65.4% | Coming soon |
| SimpleVQA | Coming soon | 61.1% | Coming soon |
| ScreenSpot Pro | Coming soon | 85.4% | Coming soon |
| ZeroBench | Coming soon | 41.0% | Coming soon |
| MedXpertQA (MM) | Coming soon | 77.1% | Coming soon |

### Knowledge

- Winner: Coming soon
- Claude Opus 5.5 public-lane score: 81.8 (Estimated · #4/189)
- GPT-5.4 public-lane score: 68.9 (Supported · #18/189)

| Benchmark | Claude Opus 5.5 | GPT-5.4 | Winner |
|-----------|-----------|-----------|--------|
| HLE w/o tools | 64.4% | 39.8% | Claude Opus 5.5 |
| HealthBench (raw) | 68.1% | Coming soon | Coming soon |
| HealthBench (length-adjusted) | 60.6% | Coming soon | Coming soon |
| HealthBench Professional | 65.6% | 48.1% | Claude Opus 5.5 |
| HealthBench Professional (raw) | 77.1% | Coming soon | Coming soon |
| BioMysteryBench (human-solvable) | 89.3% | Coming soon | Coming soon |
| BioMysteryBench (human-difficult) | 50.0% | Coming soon | Coming soon |
| SpatialBench Verified | 72.0% | Coming soon | Coming soon |
| SingleCellBench | 61.2% | Coming soon | Coming soon |
| Morphology-to-molecule matching | 34.0% | Coming soon | Coming soon |
| Medicinal chemistry | 63.5% | Coming soon | Coming soon |
| Protein Design | 60.2% | Coming soon | Coming soon |
| Protein Design library ranking | 56.0% | Coming soon | Coming soon |
| De novo protein-binder design | 82.6% | Coming soon | Coming soon |
| Protocols (troubleshooting) | 73.7% | Coming soon | Coming soon |
| Protocols (understanding) | 69.0% | Coming soon | Coming soon |
| GPQA | Coming soon | 92.8% | Coming soon |
| HLE | Coming soon | 52.1% | Coming soon |
| GPQA-D | Coming soon | 92.8% | Coming soon |
| HealthBench Hard | Coming soon | 40.1% | Coming soon |
| MedXpertQA (Text) | Coming soon | 59.6% | Coming soon |

### Multilingual

- Winner: Coming soon
- Claude Opus 5.5 public-lane score: Coming soon
- GPT-5.4 public-lane score: Coming soon

| Benchmark | Claude Opus 5.5 | GPT-5.4 | Winner |
|-----------|-----------|-----------|--------|
| GMMLU | 94.3% | Coming soon | Coming soon |
| MILU | 93.1% | Coming soon | Coming soon |

### Instruction Following

- Winner: Coming soon
- Claude Opus 5.5 public-lane score: Coming soon
- GPT-5.4 public-lane score: 89.2 (#19/124)

### Mathematics

- Winner: Coming soon
- Claude Opus 5.5 public-lane score: Coming soon
- GPT-5.4 public-lane score: 64.4 (Unranked · 2 rankable rows)

| Benchmark | Claude Opus 5.5 | GPT-5.4 | Winner |
|-----------|-----------|-----------|--------|
| ArXivMath Aug. 2026 (no tools) | 91.2% | Coming soon | Coming soon |
| ArXivMath Aug. 2026 (tools) | 96.9% | Coming soon | Coming soon |
| FrontierMath v2 (Tiers 1-3) | Coming soon | 47.600% | Coming soon |
| FrontierMath v2 (Tier 4) | Coming soon | 27.100% | Coming soon |

## FAQ

### Which is better overall, Claude Opus 5.5 or GPT-5.4?

Claude Opus 5.5 is ahead overall on BenchLM right now.

### Where is the biggest gap between Claude Opus 5.5 and GPT-5.4?

The widest category gap is in agentic, where the averages are 82.9 for Claude Opus 5.5 and 52 for GPT-5.4.

### How many benchmarks does Claude Opus 5.5 cover on BenchLM?

Claude Opus 5.5 currently has 47 sourced benchmark scores on BenchLM.

### How many benchmarks does GPT-5.4 cover on BenchLM?

GPT-5.4 currently has 38 sourced benchmark scores on BenchLM.

## Related Comparisons

- [Claude Opus 5.5 vs GPT-5.4 mini](/compare/claude-opus-5-5-vs-gpt-5-4-mini)
- [GPT-5.4 vs GPT-5.4 mini](/compare/gpt-5-4-vs-gpt-5-4-mini)
- [Claude Opus 5.5 vs GPT-5.4 Pro](/compare/claude-opus-5-5-vs-gpt-5-4-pro)
- [GPT-5.4 vs GPT-5.4 Pro](/compare/gpt-5-4-vs-gpt-5-4-pro)

## Explore More

- [Claude Opus 5.5 profile](/models/claude-opus-5-5)
- [GPT-5.4 profile](/models/gpt-5-4)
- [Compare Pricing](/llm-pricing)
- [Alternative Finder](/tools/alternative-finder)
- [LLM Selector](/tools/llm-selector)
- [Overall Rankings](/best/overall)