Model comparison
Claude 3 Haiku vs GPT-5.4 mini
GPT-5.4 mini is the stronger model overall, scoring 45.0 to 25.9 on the Noometry Index.
Last verified . 27 shared benchmarks.
Summary
- They share 27 benchmarks with published results for both. Claude 3 Haiku scores higher in 0 categories and GPT-5.4 mini in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5.4 mini leads 45.5 to 9.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 1.8% for Claude 3 Haiku and 88.9% for GPT-5.4 mini.
Side by side
| Claude 3 Haiku | GPT-5.4 mini | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 25.9 | 45.0 |
| Released | 2024-03-07 | 2026-03-17 |
| Weights | Proprietary | Proprietary |
| Context window | — | 400K |
| Max output | — | 128K |
| Input $ / M tokens | — | $0.75 |
| Output $ / M tokens | — | $4.50 |
| Results tracked | 37 | 46 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.4 mini leads
Claude 3 Haiku: 26.4 (#325), GPT-5.4 mini: 45.2 (#72)
| Benchmark | Claude 3 Haiku | GPT-5.4 mini |
|---|---|---|
| WeirdML | 9.8% | 60.3% |
| LMArena Coding | 1199 | 1438 |
| FrontierCode | — | 27% |
| LMArena WebDev | — | 1397 |
| SciCode | — | 49.9% |
| BigCodeBench Instruct | 39.4% | — |
| BigCodeBench Complete | 50.1% | — |
| CadEval | 12% | — |
| ALE-Bench | — | 1,189 |
| HumanEval+ | 68.9% | — |
| MBPP+ | 68.8% | — |
Agentic & Tool Use Not comparable
Claude 3 Haiku: —, GPT-5.4 mini: 29.9 (#81)
| Benchmark | Claude 3 Haiku | GPT-5.4 mini |
|---|---|---|
| DeepResearch Bench | — | 36.3% |
Reasoning GPT-5.4 mini leads
Claude 3 Haiku: 16.3 (#307), GPT-5.4 mini: 30.4 (#85)
| Benchmark | Claude 3 Haiku | GPT-5.4 mini |
|---|---|---|
| Kagi LLM Benchmark | 34.2% | 37.9% |
| LMArena Hard Prompts | 1174 | 1424 |
| DTBench | 50.1% | 80% |
| LMCA | 8.8% | 40.8% |
| Epoch Capabilities Index | 118.35 | 148.84 |
| ForecastBench | 53.2 | 57 |
| ARC-AGI-2 | — | 18.9% |
| NYT Connections (extended) | — | 61.8% |
| ARC-AGI-1 | — | 63.7% |
| CritPt | — | 10% |
| Chess Puzzles | — | 24% |
| Thematic Generalization | — | 61.7% |
| Mystery Game Puzzles | — | 11% |
| WinoGrande | 74.2% | — |
Math GPT-5.4 mini leads
Claude 3 Haiku: 9.8 (#319), GPT-5.4 mini: 45.5 (#75)
| Benchmark | Claude 3 Haiku | GPT-5.4 mini |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 1.8% | 88.9% |
| LMArena Math | 1188 | 1419 |
| FrontierMath (Tiers 1-3) | — | 51.2% |
| FrontierMath Tier 4 | — | 9.8% |
| ProofBench | — | 21% |
| MATH Level 5 | 14.9% | — |
| FrontierMath (Feb 2025 set) | — | 28.3% |
| FrontierMath Tier 4 (v1) | — | 2.1% |
Knowledge GPT-5.4 mini leads
Claude 3 Haiku: 17.3 (#285), GPT-5.4 mini: 51.5 (#67)
| Benchmark | Claude 3 Haiku | GPT-5.4 mini |
|---|---|---|
| GPQA Diamond | 36.3% | 86.9% |
| LMArena Expert | 1148 | 1435 |
| SimpleQA Verified | — | 29.4% |
| Confabulations | 34.2% | — |
| Vectara Hallucination Rate | — | 5.5% |
| MMLU | 73.8% | — |
Multimodal GPT-5.4 mini leads
Claude 3 Haiku: 23.6 (#128), GPT-5.4 mini: 39.7 (#56)
| Benchmark | Claude 3 Haiku | GPT-5.4 mini |
|---|---|---|
| LMArena Vision | 950 | 1245 |
| ScienceQA | 72% | — |
Multilingual GPT-5.4 mini leads
Claude 3 Haiku: 36.0 (#243), GPT-5.4 mini: 51.9 (#96)
| Benchmark | Claude 3 Haiku | GPT-5.4 mini |
|---|---|---|
| LMArena Non-English | 1178 | 1405 |
| LMArena Chinese | 1155 | 1446 |
| LMArena French | 1195 | 1440 |
| LMArena German | 1174 | 1409 |
| LMArena Japanese | 1102 | 1374 |
| LMArena Korean | 1109 | 1368 |
| LMArena Russian | 1204 | 1417 |
| LMArena Spanish | 1166 | 1405 |
Instruction Following GPT-5.4 mini leads
Claude 3 Haiku: 61.3 (#247), GPT-5.4 mini: 74.1 (#102)
| Benchmark | Claude 3 Haiku | GPT-5.4 mini |
|---|---|---|
| LMArena Instruction Following | 1173 | 1405 |
Long Context GPT-5.4 mini leads
Claude 3 Haiku: 36.1 (#237), GPT-5.4 mini: 43.0 (#112)
| Benchmark | Claude 3 Haiku | GPT-5.4 mini |
|---|---|---|
| LMArena Longer Query | 1190 | 1407 |
Writing & Preference GPT-5.4 mini leads
Claude 3 Haiku: 29.7 (#291), GPT-5.4 mini: 64.0 (#58)
| Benchmark | Claude 3 Haiku | GPT-5.4 mini |
|---|---|---|
| LMArena Text | 1195 | 1412 |
| LMArena Creative Writing | 1157 | 1370 |
| EQ-Bench Creative Writing | 717 | 1665 |
| LMArena Multi-Turn | 1190 | 1429 |
Frequently asked questions
Is Claude 3 Haiku better than GPT-5.4 mini?
GPT-5.4 mini is the stronger model overall, scoring 45.0 to 25.9 on the Noometry Index.
Is Claude 3 Haiku or GPT-5.4 mini better for coding?
GPT-5.4 mini scores higher on coding benchmarks: 45.2 versus 26.4 in the Noometry coding category.
How many benchmarks do Claude 3 Haiku and GPT-5.4 mini share?
27 benchmarks have published results for both models. Claude 3 Haiku has 37 scored results on Noometry and GPT-5.4 mini has 46.