Model comparison
Claude Haiku 5.5 vs GPT-5.2
GPT-5.2 is the stronger model overall, scoring 54.1 to 49.5 on the Noometry Index. Claude Haiku 5.5 costs 24× less per token, which makes it the better buy when GPT-5.2's lead doesn't matter for your workload.
Last verified . 9 shared benchmarks.
Summary
- They share 9 benchmarks with published results for both. Claude Haiku 5.5 scores higher in 1 category and GPT-5.2 in 4 categories; 5 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GPT-5.2 leads 50.2 to 35.2.
- The biggest single-benchmark swing is NYT Connections (extended): 65.7% for Claude Haiku 5.5 and 83.6% for GPT-5.2.
- Claude Haiku 5.5 is cheaper at $0.10 / $0.50 per million input/output tokens, against $1.75 / $14 for GPT-5.2.
- Claude Haiku 5.5 accepts more context: 1M tokens versus 400K.
Side by side
| Claude Haiku 5.5 | GPT-5.2 | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 49.5 | 54.1 |
| Released | 2026-10-07 | 2025-12-11 |
| Weights | Proprietary | Proprietary |
| Context window | 1M | 400K |
| Max output | 128K | 128K |
| Input $ / M tokens | $0.10 | $1.75 |
| Output $ / M tokens | $0.50 | $14 |
| Results tracked | 9 | 67 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.2 leads
Claude Haiku 5.5: 49.2 (#50), GPT-5.2: 51.6 (#37)
| Benchmark | Claude Haiku 5.5 | GPT-5.2 |
|---|---|---|
| LMArena WebDev | 1587 | 1416 |
| SWE-bench Verified | — | 73.8% |
| SWE-bench Verified (bash only) | — | 72.8% |
| SWE-bench Multilingual | — | 66.7% |
| GSO | — | 27.4% |
| WeirdML | — | 72.2% |
| LMArena Coding | — | 1447 |
| ALE-Bench | — | 1,294 |
| AlgoTune | — | 2.05 |
Agentic & Tool Use Not comparable
Claude Haiku 5.5: —, GPT-5.2: 40.2 (#24)
| Benchmark | Claude Haiku 5.5 | GPT-5.2 |
|---|---|---|
| Terminal-Bench | — | 64.9% |
| Berkeley Function Calling Leaderboard | — | 55.9% |
| GDPval | — | 49.7% |
| Remote Labor Index | — | 2.5% |
| τ²-bench Airline | — | 83% |
| τ²-bench Banking | — | 32.2% |
| τ²-bench Retail | — | 81.6% |
| τ²-bench Telecom | — | 89.7% |
| DeepResearch Bench | — | 41.1% |
| LMArena Search | — | 1207 |
| METR Time Horizons | — | 75.3% |
| Vending-Bench 2 | — | 3,591 |
Reasoning GPT-5.2 leads
Claude Haiku 5.5: 35.2 (#69), GPT-5.2: 50.2 (#35)
| Benchmark | Claude Haiku 5.5 | GPT-5.2 |
|---|---|---|
| NYT Connections (extended) | 65.7% | 83.6% |
| Mystery Game Puzzles | 30% | 23% |
| ARC-AGI-2 | — | 52.9% |
| SimpleBench | — | 45.8% |
| Kagi LLM Benchmark | — | 73.3% |
| ARC-AGI-1 | — | 86.2% |
| Chess Puzzles | — | 49% |
| EnigmaEval | — | 10.4% |
| EBR-Bench | — | 23% |
| LMArena Hard Prompts | — | 1445 |
| DTBench | — | 90.9% |
| LMCA | — | 43.9% |
| Epoch Capabilities Index | — | 153.45 |
| ForecastBench | — | 60.1 |
Math Claude Haiku 5.5 leads
Claude Haiku 5.5: 73.6 (#18), GPT-5.2: 60.0 (#38)
| Benchmark | Claude Haiku 5.5 | GPT-5.2 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 75.1% | 67.4% |
| FrontierMath Tier 4 | 46.3% | 31.7% |
| OTIS Mock AIME 2024-2025 | 98.9% | 96.1% |
| MathArena Final-Answer Competitions | — | 72% |
| ProofBench | — | 15% |
| LMArena Math | — | 1440 |
| FrontierMath (Feb 2025 set) | — | 40.7% |
| FrontierMath Tier 4 (v1) | — | 18.8% |
Knowledge GPT-5.2 leads
Claude Haiku 5.5: 50.8 (#70), GPT-5.2: 59.3 (#32)
| Benchmark | Claude Haiku 5.5 | GPT-5.2 |
|---|---|---|
| GPQA Diamond | 89.6% | 91.4% |
| SimpleQA Verified | 23.8% | 37.1% |
| Humanity's Last Exam | — | 27.8% |
| Vectara Hallucination Rate | — | 8.4% |
| LMArena Expert | — | 1445 |
Multimodal GPT-5.2 leads
Claude Haiku 5.5: 43.7 (#21), GPT-5.2: 51.3 (#7)
| Benchmark | Claude Haiku 5.5 | GPT-5.2 |
|---|---|---|
| Furniture Assembly | 47.5% | 38.3% |
| LMArena Vision | — | 1268 |
| VPCT | — | 84% |
| LMArena Document | — | 1405 |
Multilingual Not comparable
Claude Haiku 5.5: —, GPT-5.2: 53.4 (#67)
| Benchmark | Claude Haiku 5.5 | GPT-5.2 |
|---|---|---|
| LMArena Non-English | — | 1425 |
| LMArena Chinese | — | 1460 |
| LMArena French | — | 1455 |
| LMArena German | — | 1448 |
| LMArena Japanese | — | 1420 |
| LMArena Korean | — | 1392 |
| LMArena Russian | — | 1440 |
| LMArena Spanish | — | 1433 |
Instruction Following Not comparable
Claude Haiku 5.5: —, GPT-5.2: 74.7 (#89)
| Benchmark | Claude Haiku 5.5 | GPT-5.2 |
|---|---|---|
| LMArena Instruction Following | — | 1417 |
Long Context Not comparable
Claude Haiku 5.5: —, GPT-5.2: 44.0 (#78)
| Benchmark | Claude Haiku 5.5 | GPT-5.2 |
|---|---|---|
| CL-bench | — | 18.2% |
| LMArena Longer Query | — | 1428 |
Writing & Preference Not comparable
Claude Haiku 5.5: —, GPT-5.2: 66.8 (#32)
| Benchmark | Claude Haiku 5.5 | GPT-5.2 |
|---|---|---|
| LMArena Text | — | 1439 |
| LMArena Creative Writing | — | 1401 |
| EQ-Bench Creative Writing | — | 1703 |
| LMArena Multi-Turn | — | 1458 |
Frequently asked questions
Is Claude Haiku 5.5 better than GPT-5.2?
GPT-5.2 is the stronger model overall, scoring 54.1 to 49.5 on the Noometry Index. Claude Haiku 5.5 costs 24× less per token, which makes it the better buy when GPT-5.2's lead doesn't matter for your workload.
Which is cheaper, Claude Haiku 5.5 or GPT-5.2?
Claude Haiku 5.5 is cheaper. It lists at $0.10 per million input tokens and $0.50 per million output tokens; GPT-5.2 lists at $1.75 and $14.
Is Claude Haiku 5.5 or GPT-5.2 better for coding?
GPT-5.2 scores higher on coding benchmarks: 51.6 versus 49.2 in the Noometry coding category.
Which has the bigger context window?
Claude Haiku 5.5 does, with 1M tokens against 400K.
How many benchmarks do Claude Haiku 5.5 and GPT-5.2 share?
9 benchmarks have published results for both models. Claude Haiku 5.5 has 9 scored results on Noometry and GPT-5.2 has 67.