Model comparison
Claude Haiku 4.5 vs GPT-4.1
Claude Haiku 4.5 is the stronger model overall, scoring 39.3 to 35.7 on the Noometry Index.
Last verified . 40 shared benchmarks.
Summary
- They share 40 benchmarks with published results for both. Claude Haiku 4.5 scores higher in 8 categories and GPT-4.1 in 2 categories; 5 gaps are clear of the uncertainty.
- The widest gap is in math, where Claude Haiku 4.5 leads 43.0 to 21.4.
- The biggest single-benchmark swing is ARC-AGI-1: 47.7% for Claude Haiku 4.5 and 5.5% for GPT-4.1.
- Claude Haiku 4.5 is cheaper at $1 / $5 per million input/output tokens, against $2 / $8 for GPT-4.1.
- GPT-4.1 accepts more context: 1.05M tokens versus 200K.
Side by side
| Claude Haiku 4.5 | GPT-4.1 | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 39.3 | 35.7 |
| Released | 2025-10-15 | 2025-04-14 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | 1.05M |
| Max output | 64K | 33K |
| Input $ / M tokens | $1 | $2 |
| Output $ / M tokens | $5 | $8 |
| Results tracked | 53 | 52 |
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: 43.9 (#71), GPT-4.1: 34.2 (#228)
| Benchmark | Claude Haiku 4.5 | GPT-4.1 |
|---|---|---|
| SWE-bench Verified (bash only) | 66.6% | 39.6% |
| WeirdML | 45.4% | 39% |
| LMArena Coding | 1454 | 1391 |
| ALE-Bench | 653.48 | 558.1 |
| SWE-bench Verified | — | 48.5% |
| Aider Polyglot | — | 52.4% |
| LMArena WebDev | 1330 | — |
| SWE-bench Multilingual | 64.7% | — |
| SciCode | 43.3% | — |
| CadEval | — | 42% |
Agentic & Tool Use Too close to call
Claude Haiku 4.5: 33.5 (#50), GPT-4.1: 34.4 (#40)
| Benchmark | Claude Haiku 4.5 | GPT-4.1 |
|---|---|---|
| Berkeley Function Calling Leaderboard | 68.7% | 54% |
| Terminal-Bench | 35.5% | — |
| DeepResearch Bench | 45.5% | — |
| BALROG | 31.2% | — |
| ExploitBench | 13.7% | — |
| Vending-Bench 2 | 458.89 | — |
Reasoning Claude Haiku 4.5 leads
Claude Haiku 4.5: 15.1 (#310), GPT-4.1: 11.7 (#334)
| Benchmark | Claude Haiku 4.5 | GPT-4.1 |
|---|---|---|
| ARC-AGI-2 | 4% | 0.4% |
| ARC-AGI-1 | 47.7% | 5.5% |
| Chess Puzzles | 8% | 6% |
| LMArena Hard Prompts | 1420 | 1384 |
| DTBench | 73.6% | 68.3% |
| LMCA | 30.9% | 25.6% |
| Epoch Capabilities Index | 142.42 | 136.79 |
| ForecastBench | 61.4 | 61.5 |
| SimpleBench | — | 27% |
| Kagi LLM Benchmark | — | 52.3% |
| NYT Connections (extended) | 14.3% | — |
| CritPt | 0% | — |
| EnigmaEval | — | 2.2% |
Math Claude Haiku 4.5 leads
Claude Haiku 4.5: 43.0 (#69), GPT-4.1: 21.4 (#269)
| Benchmark | Claude Haiku 4.5 | GPT-4.1 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.7% | 38.3% |
| Omni-MATH | 56.1% | 47.1% |
| LMArena Math | 1395 | 1369 |
| MATH Level 5 | 96.4% | 83% |
| FrontierMath (Feb 2025 set) | 5.9% | 5.5% |
| FrontierMath Tier 4 (v1) | 2.1% | 0% |
| FrontierMath (Tiers 1-3) | — | 6% |
Knowledge Too close to call
Claude Haiku 4.5: 37.6 (#146), GPT-4.1: 37.0 (#152)
| Benchmark | Claude Haiku 4.5 | GPT-4.1 |
|---|---|---|
| GPQA Diamond | 71.2% | 66.9% |
| SimpleQA Verified | 13.2% | 31.1% |
| MMLU-Pro | 77.7% | 81.1% |
| Vectara Hallucination Rate | 9.8% | 5.6% |
| GPQA (HELM) | 60.5% | 65.9% |
| LMArena Expert | 1443 | 1364 |
| Humanity's Last Exam | — | 5.4% |
Multimodal GPT-4.1 leads
Claude Haiku 4.5: 27.0 (#111), GPT-4.1: 37.5 (#66)
| Benchmark | Claude Haiku 4.5 | GPT-4.1 |
|---|---|---|
| LMArena Vision | — | 1210 |
| GeoBench | — | 72% |
| Blueprint-Bench 2 | 0% | — |
| LMArena Document | 1420 | — |
Multilingual Too close to call
Claude Haiku 4.5: 50.1 (#124), GPT-4.1: 49.6 (#128)
| Benchmark | Claude Haiku 4.5 | GPT-4.1 |
|---|---|---|
| LMArena Non-English | 1376 | 1370 |
| LMArena Chinese | 1415 | 1382 |
| LMArena French | 1408 | 1383 |
| LMArena German | 1377 | 1380 |
| LMArena Japanese | 1339 | 1320 |
| LMArena Korean | 1346 | 1339 |
| LMArena Russian | 1380 | 1377 |
| LMArena Spanish | 1420 | 1375 |
Instruction Following Too close to call
Claude Haiku 4.5: 71.6 (#144), GPT-4.1: 71.5 (#148)
| Benchmark | Claude Haiku 4.5 | GPT-4.1 |
|---|---|---|
| IFEval | 80.1% | 83.8% |
| LMArena Instruction Following | 1414 | 1367 |
Long Context Claude Haiku 4.5 leads
Claude Haiku 4.5: 43.6 (#86), GPT-4.1: 40.0 (#158)
| Benchmark | Claude Haiku 4.5 | GPT-4.1 |
|---|---|---|
| LMArena Longer Query | 1427 | 1385 |
| Fiction.LiveBench | — | 63.9% |
Writing & Preference Too close to call
Claude Haiku 4.5: 58.1 (#119), GPT-4.1: 57.8 (#120)
| Benchmark | Claude Haiku 4.5 | GPT-4.1 |
|---|---|---|
| LMArena Text | 1396 | 1383 |
| LMArena Creative Writing | 1371 | 1363 |
| WildBench | 83.9% | 85.4% |
| LMArena Multi-Turn | 1409 | 1398 |
| EQ-Bench Creative Writing | — | 1420 |
| EQ-Bench 4 | 1064 | — |
Frequently asked questions
Is Claude Haiku 4.5 better than GPT-4.1?
Claude Haiku 4.5 is the stronger model overall, scoring 39.3 to 35.7 on the Noometry Index.
Which is cheaper, Claude Haiku 4.5 or GPT-4.1?
Claude Haiku 4.5 is cheaper. It lists at $1 per million input tokens and $5 per million output tokens; GPT-4.1 lists at $2 and $8.
Is Claude Haiku 4.5 or GPT-4.1 better for coding?
Claude Haiku 4.5 scores higher on coding benchmarks: 43.9 versus 34.2 in the Noometry coding category.
Which has the bigger context window?
GPT-4.1 does, with 1.05M tokens against 200K.
How many benchmarks do Claude Haiku 4.5 and GPT-4.1 share?
40 benchmarks have published results for both models. Claude Haiku 4.5 has 53 scored results on Noometry and GPT-4.1 has 52.