Model comparison
Claude Haiku 4.5 vs GPT-5
GPT-5 is the stronger model overall, scoring 50.9 to 39.5 on the Noometry Index. Claude Haiku 4.5 costs 1.7× less per token, which makes it the better buy when GPT-5's lead doesn't matter for your workload.
Last verified . 45 shared benchmarks.
Summary
- They share 45 benchmarks with published results for both. Claude Haiku 4.5 scores higher in 1 category and GPT-5 in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in long context, where GPT-5 leads 69.5 to 43.6.
- The biggest single-benchmark swing is SimpleQA Verified: 13.2% for Claude Haiku 4.5 and 50.1% for GPT-5.
- Claude Haiku 4.5 is cheaper at $1 / $5 per million input/output tokens, against $1.25 / $10 for GPT-5.
- GPT-5 accepts more context: 400K tokens versus 200K.
Side by side
| Claude Haiku 4.5 | GPT-5 | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 39.5 | 50.9 |
| Released | 2025-10-15 | 2025-08-07 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | 400K |
| Max output | 64K | 128K |
| Input $ / M tokens | $1 | $1.25 |
| Output $ / M tokens | $5 | $10 |
| Results tracked | 53 | 69 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5 leads
Claude Haiku 4.5: 44.0 (#78), GPT-5: 50.3 (#47)
| Benchmark | Claude Haiku 4.5 | GPT-5 |
|---|---|---|
| SWE-bench Verified (bash only) | 66.6% | 65% |
| LMArena WebDev | 1330 | 1418 |
| SciCode | 43.3% | 42.9% |
| WeirdML | 45.4% | 60.7% |
| LMArena Coding | 1453 | 1436 |
| ALE-Bench | 653.48 | 1,162 |
| SWE-bench Verified | — | 73.6% |
| Aider Polyglot | — | 88% |
| SWE-bench Multilingual | 64.7% | — |
| GSO | — | 6.9% |
| AlgoTune | — | 1.67 |
Agentic & Tool Use Too close to call
Claude Haiku 4.5: 33.6 (#52), GPT-5: 33.1 (#56)
| Benchmark | Claude Haiku 4.5 | GPT-5 |
|---|---|---|
| Terminal-Bench | 35.5% | 49.6% |
| DeepResearch Bench | 45.5% | 49.6% |
| BALROG | 31.2% | 32.8% |
| Berkeley Function Calling Leaderboard | 68.7% | — |
| GDPval | — | 34.8% |
| Remote Labor Index | — | 1.7% |
| ExploitBench | 13.7% | — |
| LMArena Search | — | 1133 |
| METR Time Horizons | — | 69.6% |
| Vending-Bench 2 | 458.89 | — |
Reasoning GPT-5 leads
Claude Haiku 4.5: 15.1 (#320), GPT-5: 38.3 (#64)
| Benchmark | Claude Haiku 4.5 | GPT-5 |
|---|---|---|
| ARC-AGI-2 | 4% | 9.9% |
| ARC-AGI-1 | 47.7% | 65.7% |
| CritPt | 0% | 12.6% |
| Chess Puzzles | 8% | 37% |
| LMArena Hard Prompts | 1420 | 1416 |
| DTBench | 73.6% | 90.7% |
| LMCA | 30.9% | 40% |
| Epoch Capabilities Index | 142.41 | 150 |
| ForecastBench | 61.4 | 61.4 |
| SimpleBench | — | 56.7% |
| Kagi LLM Benchmark | — | 72.7% |
| NYT Connections (extended) | 14.3% | — |
| EnigmaEval | — | 10.5% |
| EBR-Bench | — | 12.7% |
| Mystery Game Puzzles | — | 23% |
Math GPT-5 leads
Claude Haiku 4.5: 44.9 (#78), GPT-5: 55.0 (#44)
| Benchmark | Claude Haiku 4.5 | GPT-5 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.7% | 91.4% |
| Omni-MATH | 56.1% | 64.7% |
| LMArena Math | 1396 | 1407 |
| MATH Level 5 | 96.4% | 98.1% |
| FrontierMath (Feb 2025 set) | 5.9% | 32.4% |
| FrontierMath Tier 4 (v1) | 2.1% | 12.5% |
| FrontierMath (Tiers 1-3) | — | 55.4% |
| FrontierMath Tier 4 | — | 22% |
| ProofBench | — | 18% |
Knowledge GPT-5 leads
Claude Haiku 4.5: 37.7 (#153), GPT-5: 56.6 (#43)
| Benchmark | Claude Haiku 4.5 | GPT-5 |
|---|---|---|
| GPQA Diamond | 71.2% | 86.2% |
| SimpleQA Verified | 13.2% | 50.1% |
| MMLU-Pro | 77.7% | 86.3% |
| Vectara Hallucination Rate | 9.8% | 14.7% |
| GPQA (HELM) | 60.5% | 79.2% |
| LMArena Expert | 1442 | 1419 |
| Humanity's Last Exam | — | 25.3% |
| Confabulations | — | 10.3% |
Multimodal GPT-5 leads
Claude Haiku 4.5: 26.8 (#118), GPT-5: 46.8 (#13)
| Benchmark | Claude Haiku 4.5 | GPT-5 |
|---|---|---|
| LMArena Vision | — | 1232 |
| GeoBench | — | 81% |
| VPCT | — | 66% |
| Blueprint-Bench 2 | 0% | — |
| LMArena Document | 1420 | — |
Multilingual GPT-5 leads
Claude Haiku 4.5: 49.9 (#129), GPT-5: 51.4 (#110)
| Benchmark | Claude Haiku 4.5 | GPT-5 |
|---|---|---|
| LMArena Non-English | 1377 | 1397 |
| LMArena Chinese | 1417 | 1422 |
| LMArena French | 1408 | 1410 |
| LMArena German | 1375 | 1416 |
| LMArena Japanese | 1339 | 1409 |
| LMArena Korean | 1347 | 1360 |
| LMArena Russian | 1381 | 1406 |
| LMArena Spanish | 1420 | 1399 |
Instruction Following GPT-5 leads
Claude Haiku 4.5: 71.4 (#149), GPT-5: 73.8 (#113)
| Benchmark | Claude Haiku 4.5 | GPT-5 |
|---|---|---|
| IFEval | 80.1% | 87.5% |
| LMArena Instruction Following | 1414 | 1388 |
Long Context GPT-5 leads
Claude Haiku 4.5: 43.6 (#92), GPT-5: 69.5 (#2)
| Benchmark | Claude Haiku 4.5 | GPT-5 |
|---|---|---|
| LMArena Longer Query | 1427 | 1399 |
| Fiction.LiveBench | — | 97.2% |
Writing & Preference GPT-5 leads
Claude Haiku 4.5: 57.9 (#123), GPT-5: 63.4 (#65)
| Benchmark | Claude Haiku 4.5 | GPT-5 |
|---|---|---|
| LMArena Text | 1396 | 1406 |
| LMArena Creative Writing | 1372 | 1365 |
| WildBench | 83.9% | 85.7% |
| LMArena Multi-Turn | 1409 | 1426 |
| Short-Story Creative Writing | — | 86% |
| EQ-Bench Creative Writing | — | 1627 |
| EQ-Bench 4 | 1064 | — |
Frequently asked questions
Is Claude Haiku 4.5 better than GPT-5?
GPT-5 is the stronger model overall, scoring 50.9 to 39.5 on the Noometry Index. Claude Haiku 4.5 costs 1.7× less per token, which makes it the better buy when GPT-5's lead doesn't matter for your workload.
Which is cheaper, Claude Haiku 4.5 or GPT-5?
Claude Haiku 4.5 is cheaper. It lists at $1 per million input tokens and $5 per million output tokens; GPT-5 lists at $1.25 and $10.
Is Claude Haiku 4.5 or GPT-5 better for coding?
GPT-5 scores higher on coding benchmarks: 50.3 versus 44.0 in the Noometry coding category.
Which has the bigger context window?
GPT-5 does, with 400K tokens against 200K.
How many benchmarks do Claude Haiku 4.5 and GPT-5 share?
45 benchmarks have published results for both models. Claude Haiku 4.5 has 53 scored results on Noometry and GPT-5 has 69.