Model comparison
Claude 3 Haiku vs Claude Opus 4.8
Claude Opus 4.8 is the stronger model overall, scoring 60.7 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 Claude Opus 4.8 in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where Claude Opus 4.8 leads 78.4 to 9.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 1.8% for Claude 3 Haiku and 98.3% for Claude Opus 4.8.
Side by side
| Claude 3 Haiku | Claude Opus 4.8 | |
|---|---|---|
| Provider | Anthropic | Anthropic |
| Noometry Index | 25.9 | 60.7 |
| Released | 2024-03-07 | 2026-05-28 |
| Weights | Proprietary | Proprietary |
| Context window | — | 1M |
| Max output | — | 128K |
| Input $ / M tokens | — | $5 |
| Output $ / M tokens | — | $25 |
| Results tracked | 37 | 65 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Opus 4.8 leads
Claude 3 Haiku: 26.4 (#325), Claude Opus 4.8: 59.9 (#12)
| Benchmark | Claude 3 Haiku | Claude Opus 4.8 |
|---|---|---|
| WeirdML | 9.8% | 82.9% |
| LMArena Coding | 1199 | 1490 |
| DeepSWE | — | 59% |
| FrontierCode | — | 46.5% |
| LMArena WebDev | — | 1556 |
| SciCode | — | 53.5% |
| GSO | — | 47.1% |
| BigCodeBench Instruct | 39.4% | — |
| BigCodeBench Complete | 50.1% | — |
| CadEval | 12% | — |
| ALE-Bench | — | 1,564 |
| HumanEval+ | 68.9% | — |
| MBPP+ | 68.8% | — |
Agentic & Tool Use Not comparable
Claude 3 Haiku: —, Claude Opus 4.8: 47.6 (#11)
| Benchmark | Claude 3 Haiku | Claude Opus 4.8 |
|---|---|---|
| APEX-Agents | — | 48.9% |
| OSWorld 2.0 | — | 20.6% |
| Remote Labor Index | — | 8.3% |
| τ²-bench Banking | — | 39.7% |
| DeepResearch Bench | — | 50.2% |
| PostTrainBench | — | 33.8% |
| GBAEval | — | 70.9% |
| GDP.pdf | — | 24% |
| LMArena Search | — | 1204 |
| Vending-Bench 2 | — | 5,787 |
Reasoning Claude Opus 4.8 leads
Claude 3 Haiku: 16.3 (#307), Claude Opus 4.8: 64.7 (#16)
| Benchmark | Claude 3 Haiku | Claude Opus 4.8 |
|---|---|---|
| Kagi LLM Benchmark | 34.2% | 88.8% |
| LMArena Hard Prompts | 1174 | 1482 |
| DTBench | 50.1% | 94.9% |
| LMCA | 8.8% | 57.5% |
| Epoch Capabilities Index | 118.35 | 158.21 |
| ForecastBench | 53.2 | 59.9 |
| ARC-AGI-2 | — | 72.1% |
| SimpleBench | — | 64.8% |
| NYT Connections (extended) | — | 91.1% |
| ARC-AGI-1 | — | 92.5% |
| CritPt | — | 20.9% |
| Chess Puzzles | — | 34% |
| EnigmaEval | — | 23.5% |
| EBR-Bench | — | 28.6% |
| Mystery Game Puzzles | — | 36% |
| Surface Evolver Bench | — | 87.5% |
| Bench to the Future 3 | — | 0.14 |
| WinoGrande | 74.2% | — |
Math Claude Opus 4.8 leads
Claude 3 Haiku: 9.8 (#319), Claude Opus 4.8: 78.4 (#13)
| Benchmark | Claude 3 Haiku | Claude Opus 4.8 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 1.8% | 98.3% |
| LMArena Math | 1188 | 1487 |
| FrontierMath (Tiers 1-3) | — | 80% |
| FrontierMath Tier 4 | — | 56.1% |
| MathArena Final-Answer Competitions | — | 91.8% |
| ProofBench | — | 69% |
| MATH Level 5 | 14.9% | — |
| FrontierMath (Feb 2025 set) | — | 47.2% |
| FrontierMath Tier 4 (v1) | — | 31.3% |
Knowledge Claude Opus 4.8 leads
Claude 3 Haiku: 17.3 (#285), Claude Opus 4.8: 61.3 (#29)
| Benchmark | Claude 3 Haiku | Claude Opus 4.8 |
|---|---|---|
| GPQA Diamond | 36.3% | 91% |
| LMArena Expert | 1148 | 1502 |
| SimpleQA Verified | — | 53% |
| Confabulations | 34.2% | — |
| MMLU | 73.8% | — |
Multimodal Claude Opus 4.8 leads
Claude 3 Haiku: 23.6 (#128), Claude Opus 4.8: 42.9 (#26)
| Benchmark | Claude 3 Haiku | Claude Opus 4.8 |
|---|---|---|
| LMArena Vision | 950 | 1294 |
| Blueprint-Bench 2 | — | 14.5% |
| Furniture Assembly | — | 42.5% |
| LMArena Document | — | 1475 |
| ScienceQA | 72% | — |
Multilingual Claude Opus 4.8 leads
Claude 3 Haiku: 36.0 (#243), Claude Opus 4.8: 55.2 (#33)
| Benchmark | Claude 3 Haiku | Claude Opus 4.8 |
|---|---|---|
| LMArena Non-English | 1178 | 1450 |
| LMArena Chinese | 1155 | 1507 |
| LMArena French | 1195 | 1481 |
| LMArena German | 1174 | 1472 |
| LMArena Japanese | 1102 | 1440 |
| LMArena Korean | 1109 | 1432 |
| LMArena Russian | 1204 | 1474 |
| LMArena Spanish | 1166 | 1466 |
Instruction Following Claude Opus 4.8 leads
Claude 3 Haiku: 61.3 (#247), Claude Opus 4.8: 77.4 (#24)
| Benchmark | Claude 3 Haiku | Claude Opus 4.8 |
|---|---|---|
| LMArena Instruction Following | 1173 | 1476 |
Long Context Claude Opus 4.8 leads
Claude 3 Haiku: 36.1 (#237), Claude Opus 4.8: 45.4 (#35)
| Benchmark | Claude 3 Haiku | Claude Opus 4.8 |
|---|---|---|
| LMArena Longer Query | 1190 | 1483 |
Writing & Preference Claude Opus 4.8 leads
Claude 3 Haiku: 29.7 (#291), Claude Opus 4.8: 72.0 (#16)
| Benchmark | Claude 3 Haiku | Claude Opus 4.8 |
|---|---|---|
| LMArena Text | 1195 | 1461 |
| LMArena Creative Writing | 1157 | 1454 |
| EQ-Bench Creative Writing | 717 | 1840 |
| LMArena Multi-Turn | 1190 | 1476 |
| EQ-Bench 4 | — | 1281 |
Frequently asked questions
Is Claude 3 Haiku better than Claude Opus 4.8?
Claude Opus 4.8 is the stronger model overall, scoring 60.7 to 25.9 on the Noometry Index.
Is Claude 3 Haiku or Claude Opus 4.8 better for coding?
Claude Opus 4.8 scores higher on coding benchmarks: 59.9 versus 26.4 in the Noometry coding category.
How many benchmarks do Claude 3 Haiku and Claude Opus 4.8 share?
27 benchmarks have published results for both models. Claude 3 Haiku has 37 scored results on Noometry and Claude Opus 4.8 has 65.