AI BENCHY Compare
Vergeleken modellen
Vergelijken:
Benchmarks gegenereerd uit AI BENCHY-testsuites op: 2026-03-05
| Metriek | OpenAI: GPT-5.4 medium Releasedatum: 2026-03-05 | OpenAI: GPT-5.3-Codex medium Releasedatum: 2026-02-05 | OpenAI: GPT-5.2 medium Releasedatum: 2025-12-11 |
|---|---|---|---|
| Rang | #7 | #4 | #26 |
| Gem. score | 82 | 87 | 69 |
| Consistentie | 89 | 90 | 78 |
| Kosten per resultaat | 6.533 | 4.418 | 3.057 |
| Totale kosten | $0.784 | $0.531 | $0.306 |
| Correcte tests | |||
| Slaagpercentage per poging | 86.7% | 88.9% | 80.0% |
| Instabiele tests | 2 | 2 | 4 |
| Uitvoer-tokens | 1,611 | 1,577 | 2,058 |
| Redeneer-tokens | 46,321 | 33,017 | 16,542 |
Score vs totale kosten
Categorie-uitsplitsing
| Anti-AI-trucs | Score | Consistentie | Slaagpercentage per poging | Instabiele tests | Correcte tests | Uitvoer-tokens | Redeneer-tokens |
|---|---|---|---|---|---|---|---|
| OpenAI: GPT-5.4 | 100 | 100 | 100.0% | 0 | 216 | 1,466 | |
| OpenAI: GPT-5.3-Codex | 100 | 100 | 100.0% | 0 | 216 | 1,421 | |
| OpenAI: GPT-5.2 | 70 | 73 | 77.8% | 1 | 549 | 2,002 |
| Gecombineerd | Score | Consistentie | Slaagpercentage per poging | Instabiele tests | Correcte tests | Uitvoer-tokens | Redeneer-tokens |
|---|---|---|---|---|---|---|---|
| OpenAI: GPT-5.4 | 100 | 100 | 100.0% | 0 | 301 | 3,543 | |
| OpenAI: GPT-5.3-Codex | 100 | 100 | 100.0% | 0 | 364 | 2,731 | |
| OpenAI: GPT-5.2 | 100 | 100 | 100.0% | 0 | 291 | 1,757 |
| Gegevensparsering en extractie | Score | Consistentie | Slaagpercentage per poging | Instabiele tests | Correcte tests | Uitvoer-tokens | Redeneer-tokens |
|---|---|---|---|---|---|---|---|
| OpenAI: GPT-5.4 | 99 | 100 | 100.0% | 0 | 234 | 804 | |
| OpenAI: GPT-5.3-Codex | 99 | 100 | 100.0% | 0 | 234 | 728 | |
| OpenAI: GPT-5.2 | 99 | 100 | 100.0% | 0 | 234 | 420 |
| Domeinspecifiek | Score | Consistentie | Slaagpercentage per poging | Instabiele tests | Correcte tests | Uitvoer-tokens | Redeneer-tokens |
|---|---|---|---|---|---|---|---|
| OpenAI: GPT-5.4 | 40 | 72 | 44.4% | 1 | 61 | 34,748 | |
| OpenAI: GPT-5.3-Codex | 40 | 72 | 55.6% | 1 | 64 | 25,308 | |
| OpenAI: GPT-5.2 | 40 | 72 | 55.6% | 1 | 42 | 10,342 |
| Instructies opvolgen | Score | Consistentie | Slaagpercentage per poging | Instabiele tests | Correcte tests | Uitvoer-tokens | Redeneer-tokens |
|---|---|---|---|---|---|---|---|
| OpenAI: GPT-5.4 | 100 | 100 | 100.0% | 0 | 93 | 897 | |
| OpenAI: GPT-5.3-Codex | 100 | 100 | 100.0% | 0 | 93 | 693 | |
| OpenAI: GPT-5.2 | 95 | 100 | 100.0% | 0 | 94 | 614 |
| Puzzle Solving | Score | Consistentie | Slaagpercentage per poging | Instabiele tests | Correcte tests | Uitvoer-tokens | Redeneer-tokens |
|---|---|---|---|---|---|---|---|
| OpenAI: GPT-5.4 | 70 | 72 | 88.9% | 1 | 442 | 3,832 | |
| OpenAI: GPT-5.3-Codex | 93 | 79 | 88.9% | 1 | 352 | 1,644 | |
| OpenAI: GPT-5.2 | 70 | 73 | 77.8% | 1 | 609 | 938 |
| Toolaanroepen | Score | Consistentie | Slaagpercentage per poging | Instabiele tests | Correcte tests | Uitvoer-tokens | Redeneer-tokens |
|---|---|---|---|---|---|---|---|
| OpenAI: GPT-5.4 | 100 | 100 | 100.0% | 0 | 264 | 1,031 | |
| OpenAI: GPT-5.3-Codex | 100 | 100 | 100.0% | 0 | 254 | 492 | |
| OpenAI: GPT-5.2 | 100 | 16 | 66.7% | 1 | 239 | 469 |
Snelle vergelijking
Vergelijkingspaar wisselen
Claude Sonnet 4.6nonevsGPT-5.2mediumGemini 3 Flash PreviewlowvsGPT-5.4mediumGemini 3.1 Flash Lite PreviewhighvsGPT-5.4mediumGemini 3 Flash PreviewnonevsGPT-5.2mediumGPT-5.2mediumvsQwen3.5 Plus 2026-02-15noneGemini 3.1 Flash Lite PreviewnonevsGPT-5.2mediumGemini 3 Flash PreviewlowvsGPT-5.3-CodexmediumGemini 3.1 Flash Lite PreviewlowvsGPT-5.4mediumGemini 3.1 Flash Lite PreviewhighvsGPT-5.3-CodexmediumGemini 3.1 Flash Lite PreviewlowvsGPT-5.2mediumGemini 3.1 Flash Lite PreviewnonevsGPT-5.4mediumGPT-5.2mediumvsGLM 5none