Model comparison
Claude Haiku 4.5 vs GPT-4o
Claude Haiku 4.5 is the stronger model overall, scoring 39.5 to 28.6 on the Noometry Index.
Last verified . 39 shared benchmarks.
Summary
- They share 39 benchmarks with published results for both. Claude Haiku 4.5 scores higher in 9 categories and GPT-4o in 1 category; 10 gaps are clear of the uncertainty.
- The widest gap is in math, where Claude Haiku 4.5 leads 44.9 to 10.6.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 66.7% for Claude Haiku 4.5 and 6.4% for GPT-4o.
- Claude Haiku 4.5 is cheaper at $1 / $5 per million input/output tokens, against $2.50 / $10 for GPT-4o.
- Claude Haiku 4.5 accepts more context: 200K tokens versus 128K.
Side by side
| Claude Haiku 4.5 | GPT-4o | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 39.5 | 28.6 |
| Released | 2025-10-15 | 2024-05-13 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | 128K |
| Max output | 64K | 16K |
| Input $ / M tokens | $1 | $2.50 |
| Output $ / M tokens | $5 | $10 |
| Results tracked | 53 | 72 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Haiku 4.5 leads
Claude Haiku 4.5: 44.0 (#78), GPT-4o: 24.8 (#328)
| Benchmark | Claude Haiku 4.5 | GPT-4o |
|---|---|---|
| SWE-bench Verified (bash only) | 66.6% | 21.6% |
| WeirdML | 45.4% | 25.1% |
| LMArena Coding | 1453 | 1297 |
| SWE-bench Verified | — | 31% |
| Aider Polyglot | — | 45.3% |
| LMArena WebDev | 1330 | — |
| SWE-bench Multilingual | 64.7% | — |
| SciCode | 43.3% | — |
| GSO | — | 0% |
| BigCodeBench Instruct | — | 51.1% |
| LiveBench Coding | — | 51.4% |
| BigCodeBench Complete | — | 61.1% |
| CadEval | — | 26% |
| ALE-Bench | 653.48 | — |
| HumanEval+ | — | 87.2% |
| MBPP+ | — | 72.2% |
Agentic & Tool Use Claude Haiku 4.5 leads
Claude Haiku 4.5: 33.6 (#52), GPT-4o: 21.0 (#141)
| Benchmark | Claude Haiku 4.5 | GPT-4o |
|---|---|---|
| BALROG | 31.2% | 32.3% |
| Terminal-Bench | 35.5% | — |
| Berkeley Function Calling Leaderboard | 68.7% | — |
| GDPval | — | 9.9% |
| TheAgentCompany | — | 8.6% |
| Cybench | — | 12.5% |
| DeepResearch Bench | 45.5% | — |
| ExploitBench | 13.7% | — |
| LMArena Search | — | 1006 |
| METR Time Horizons | — | 40.8% |
| Vending-Bench 2 | 458.89 | — |
Reasoning Claude Haiku 4.5 leads
Claude Haiku 4.5: 15.1 (#320), GPT-4o: 9.4 (#343)
| Benchmark | Claude Haiku 4.5 | GPT-4o |
|---|---|---|
| ARC-AGI-2 | 4% | 0% |
| ARC-AGI-1 | 47.7% | 4.5% |
| CritPt | 0% | 0% |
| Chess Puzzles | 8% | 13% |
| LMArena Hard Prompts | 1420 | 1281 |
| DTBench | 73.6% | 64.5% |
| LMCA | 30.9% | 16.6% |
| Epoch Capabilities Index | 142.41 | 128.97 |
| ForecastBench | 61.4 | 57.7 |
| SimpleBench | — | 17.8% |
| NYT Connections (extended) | 14.3% | — |
| EnigmaEval | — | 0.8% |
| LiveBench Reasoning | — | 55.8% |
| LiveBench Data Analysis | — | 60.9% |
| LiveBench | — | 55.3% |
Math Claude Haiku 4.5 leads
Claude Haiku 4.5: 44.9 (#78), GPT-4o: 10.6 (#312)
| Benchmark | Claude Haiku 4.5 | GPT-4o |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.7% | 6.4% |
| Omni-MATH | 56.1% | 29.3% |
| LMArena Math | 1396 | 1285 |
| MATH Level 5 | 96.4% | 53.3% |
| FrontierMath (Feb 2025 set) | 5.9% | 0.3% |
| FrontierMath (Tiers 1-3) | — | 0.4% |
| LiveBench Math | — | 49.5% |
| FrontierMath Tier 4 (v1) | 2.1% | — |
Knowledge Claude Haiku 4.5 leads
Claude Haiku 4.5: 37.7 (#153), GPT-4o: 28.8 (#242)
| Benchmark | Claude Haiku 4.5 | GPT-4o |
|---|---|---|
| GPQA Diamond | 71.2% | 49.2% |
| SimpleQA Verified | 13.2% | 26% |
| MMLU-Pro | 77.7% | 71.3% |
| Vectara Hallucination Rate | 9.8% | 9.6% |
| GPQA (HELM) | 60.5% | 52% |
| LMArena Expert | 1442 | 1250 |
| Humanity's Last Exam | — | 2.7% |
| Confabulations | — | 15.3% |
| MMLU | — | 88.1% |
Multimodal GPT-4o leads
Claude Haiku 4.5: 26.8 (#118), GPT-4o: 34.5 (#91)
| Benchmark | Claude Haiku 4.5 | GPT-4o |
|---|---|---|
| LMArena Vision | — | 1137 |
| Video-MME | — | 71.9% |
| GeoBench | — | 71% |
| VPCT | — | 40% |
| Blueprint-Bench 2 | 0% | — |
| LMArena Document | 1420 | — |
| ScienceQA | — | 88.5% |
Multilingual Claude Haiku 4.5 leads
Claude Haiku 4.5: 49.9 (#129), GPT-4o: 43.2 (#186)
| Benchmark | Claude Haiku 4.5 | GPT-4o |
|---|---|---|
| LMArena Non-English | 1377 | 1283 |
| LMArena Chinese | 1417 | 1277 |
| LMArena French | 1408 | 1304 |
| LMArena German | 1375 | 1282 |
| LMArena Japanese | 1339 | 1257 |
| LMArena Korean | 1347 | 1234 |
| LMArena Russian | 1381 | 1286 |
| LMArena Spanish | 1420 | 1292 |
Instruction Following Claude Haiku 4.5 leads
Claude Haiku 4.5: 71.4 (#149), GPT-4o: 66.6 (#207)
| Benchmark | Claude Haiku 4.5 | GPT-4o |
|---|---|---|
| IFEval | 80.1% | 81.7% |
| LMArena Instruction Following | 1414 | 1278 |
| LiveBench Instruction Following | — | 68.6% |
Long Context Claude Haiku 4.5 leads
Claude Haiku 4.5: 43.6 (#92), GPT-4o: 39.4 (#179)
| Benchmark | Claude Haiku 4.5 | GPT-4o |
|---|---|---|
| LMArena Longer Query | 1427 | 1289 |
| Fiction.LiveBench | — | 66.7% |
Writing & Preference Claude Haiku 4.5 leads
Claude Haiku 4.5: 57.9 (#123), GPT-4o: 52.6 (#166)
| Benchmark | Claude Haiku 4.5 | GPT-4o |
|---|---|---|
| LMArena Text | 1396 | 1300 |
| LMArena Creative Writing | 1372 | 1292 |
| WildBench | 83.9% | 82.8% |
| LMArena Multi-Turn | 1409 | 1302 |
| Short-Story Creative Writing | — | 81.8% |
| EQ-Bench 4 | 1064 | — |
| LiveBench Language | — | 47.6% |
Frequently asked questions
Is Claude Haiku 4.5 better than GPT-4o?
Claude Haiku 4.5 is the stronger model overall, scoring 39.5 to 28.6 on the Noometry Index.
Which is cheaper, Claude Haiku 4.5 or GPT-4o?
Claude Haiku 4.5 is cheaper. It lists at $1 per million input tokens and $5 per million output tokens; GPT-4o lists at $2.50 and $10.
Is Claude Haiku 4.5 or GPT-4o better for coding?
Claude Haiku 4.5 scores higher on coding benchmarks: 44.0 versus 24.8 in the Noometry coding category.
Which has the bigger context window?
Claude Haiku 4.5 does, with 200K tokens against 128K.
How many benchmarks do Claude Haiku 4.5 and GPT-4o share?
39 benchmarks have published results for both models. Claude Haiku 4.5 has 53 scored results on Noometry and GPT-4o has 72.