Model comparison
Claude Haiku 4.5 vs GPT-5.4 mini
GPT-5.4 mini is the stronger model overall, scoring 45.0 to 39.5 on the Noometry Index.
Last verified . 37 shared benchmarks.
Summary
- They share 37 benchmarks with published results for both. Claude Haiku 4.5 scores higher in 2 categories and GPT-5.4 mini in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GPT-5.4 mini leads 30.4 to 15.1.
- The biggest single-benchmark swing is NYT Connections (extended): 14.3% for Claude Haiku 4.5 and 61.8% for GPT-5.4 mini.
- GPT-5.4 mini is cheaper at $0.75 / $4.50 per million input/output tokens, against $1 / $5 for Claude Haiku 4.5.
- GPT-5.4 mini accepts more context: 400K tokens versus 200K.
Side by side
| Claude Haiku 4.5 | GPT-5.4 mini | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 39.5 | 45.0 |
| Released | 2025-10-15 | 2026-03-17 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | 400K |
| Max output | 64K | 128K |
| Input $ / M tokens | $1 | $0.75 |
| Output $ / M tokens | $5 | $4.50 |
| Results tracked | 53 | 46 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.4 mini leads
Claude Haiku 4.5: 44.0 (#78), GPT-5.4 mini: 45.2 (#72)
| Benchmark | Claude Haiku 4.5 | GPT-5.4 mini |
|---|---|---|
| LMArena WebDev | 1330 | 1397 |
| SciCode | 43.3% | 49.9% |
| WeirdML | 45.4% | 60.3% |
| LMArena Coding | 1453 | 1438 |
| ALE-Bench | 653.48 | 1,189 |
| FrontierCode | — | 27% |
| SWE-bench Verified (bash only) | 66.6% | — |
| SWE-bench Multilingual | 64.7% | — |
Agentic & Tool Use Claude Haiku 4.5 leads
Claude Haiku 4.5: 33.6 (#52), GPT-5.4 mini: 29.9 (#81)
| Benchmark | Claude Haiku 4.5 | GPT-5.4 mini |
|---|---|---|
| DeepResearch Bench | 45.5% | 36.3% |
| Terminal-Bench | 35.5% | — |
| Berkeley Function Calling Leaderboard | 68.7% | — |
| BALROG | 31.2% | — |
| ExploitBench | 13.7% | — |
| Vending-Bench 2 | 458.89 | — |
Reasoning GPT-5.4 mini leads
Claude Haiku 4.5: 15.1 (#320), GPT-5.4 mini: 30.4 (#85)
| Benchmark | Claude Haiku 4.5 | GPT-5.4 mini |
|---|---|---|
| ARC-AGI-2 | 4% | 18.9% |
| NYT Connections (extended) | 14.3% | 61.8% |
| ARC-AGI-1 | 47.7% | 63.7% |
| CritPt | 0% | 10% |
| Chess Puzzles | 8% | 24% |
| LMArena Hard Prompts | 1420 | 1424 |
| DTBench | 73.6% | 80% |
| LMCA | 30.9% | 40.8% |
| Epoch Capabilities Index | 142.41 | 148.84 |
| ForecastBench | 61.4 | 57 |
| Kagi LLM Benchmark | — | 37.9% |
| Thematic Generalization | — | 61.7% |
| Mystery Game Puzzles | — | 11% |
Math Too close to call
Claude Haiku 4.5: 44.9 (#78), GPT-5.4 mini: 45.5 (#75)
| Benchmark | Claude Haiku 4.5 | GPT-5.4 mini |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.7% | 88.9% |
| LMArena Math | 1396 | 1419 |
| FrontierMath (Feb 2025 set) | 5.9% | 28.3% |
| FrontierMath Tier 4 (v1) | 2.1% | 2.1% |
| FrontierMath (Tiers 1-3) | — | 51.2% |
| FrontierMath Tier 4 | — | 9.8% |
| ProofBench | — | 21% |
| Omni-MATH | 56.1% | — |
| MATH Level 5 | 96.4% | — |
Knowledge GPT-5.4 mini leads
Claude Haiku 4.5: 37.7 (#153), GPT-5.4 mini: 51.5 (#67)
| Benchmark | Claude Haiku 4.5 | GPT-5.4 mini |
|---|---|---|
| GPQA Diamond | 71.2% | 86.9% |
| SimpleQA Verified | 13.2% | 29.4% |
| Vectara Hallucination Rate | 9.8% | 5.5% |
| LMArena Expert | 1442 | 1435 |
| MMLU-Pro | 77.7% | — |
| GPQA (HELM) | 60.5% | — |
Multimodal GPT-5.4 mini leads
Claude Haiku 4.5: 26.8 (#118), GPT-5.4 mini: 39.7 (#56)
| Benchmark | Claude Haiku 4.5 | GPT-5.4 mini |
|---|---|---|
| LMArena Vision | — | 1245 |
| Blueprint-Bench 2 | 0% | — |
| LMArena Document | 1420 | — |
Multilingual GPT-5.4 mini leads
Claude Haiku 4.5: 49.9 (#129), GPT-5.4 mini: 51.9 (#96)
| Benchmark | Claude Haiku 4.5 | GPT-5.4 mini |
|---|---|---|
| LMArena Non-English | 1377 | 1405 |
| LMArena Chinese | 1417 | 1446 |
| LMArena French | 1408 | 1440 |
| LMArena German | 1375 | 1409 |
| LMArena Japanese | 1339 | 1374 |
| LMArena Korean | 1347 | 1368 |
| LMArena Russian | 1381 | 1417 |
| LMArena Spanish | 1420 | 1405 |
Instruction Following GPT-5.4 mini leads
Claude Haiku 4.5: 71.4 (#149), GPT-5.4 mini: 74.1 (#102)
| Benchmark | Claude Haiku 4.5 | GPT-5.4 mini |
|---|---|---|
| LMArena Instruction Following | 1414 | 1405 |
| IFEval | 80.1% | — |
Long Context Too close to call
Claude Haiku 4.5: 43.6 (#92), GPT-5.4 mini: 43.0 (#112)
| Benchmark | Claude Haiku 4.5 | GPT-5.4 mini |
|---|---|---|
| LMArena Longer Query | 1427 | 1407 |
Writing & Preference GPT-5.4 mini leads
Claude Haiku 4.5: 57.9 (#123), GPT-5.4 mini: 64.0 (#58)
| Benchmark | Claude Haiku 4.5 | GPT-5.4 mini |
|---|---|---|
| LMArena Text | 1396 | 1412 |
| LMArena Creative Writing | 1372 | 1370 |
| LMArena Multi-Turn | 1409 | 1429 |
| EQ-Bench Creative Writing | — | 1665 |
| WildBench | 83.9% | — |
| EQ-Bench 4 | 1064 | — |
Frequently asked questions
Is Claude Haiku 4.5 better than GPT-5.4 mini?
GPT-5.4 mini is the stronger model overall, scoring 45.0 to 39.5 on the Noometry Index.
Which is cheaper, Claude Haiku 4.5 or GPT-5.4 mini?
GPT-5.4 mini is cheaper. It lists at $0.75 per million input tokens and $4.50 per million output tokens; Claude Haiku 4.5 lists at $1 and $5.
Is Claude Haiku 4.5 or GPT-5.4 mini better for coding?
GPT-5.4 mini scores higher on coding benchmarks: 45.2 versus 44.0 in the Noometry coding category.
Which has the bigger context window?
GPT-5.4 mini does, with 400K tokens against 200K.
How many benchmarks do Claude Haiku 4.5 and GPT-5.4 mini share?
37 benchmarks have published results for both models. Claude Haiku 4.5 has 53 scored results on Noometry and GPT-5.4 mini has 46.