AI BENCHY Compare
Google: Gemini 3.1 Flash Lite Preview vs OpenAI: GPT-5.4
Bandingkan:
Benchmark dihasilkan dari suite pengujian AI BENCHY pada: 2026-03-05
| Metrik | Google: Gemini 3.1 Flash Lite Preview high Rilis: 2026-03-03 | OpenAI: GPT-5.4 medium Rilis: 2026-03-05 |
|---|---|---|
| Skor Rata-rata | 8.1 | 8.2 |
| Tes benar | ||
| Peringkat | #10 | #7 |
| Konsistensi | 9.6 | 8.9 |
| Biaya per hasil | 20.937 | 6.533 |
| Total Biaya | $2.303 | $0.784 |
| Tingkat lulus per percobaan | 75.6% | 86.7% |
| Tes tidak stabil | 1 | 2 |
| common.totalAttempts | 45 (15 x 3) | 45 (15 x 3) |
| Token output | 1,166 | 1,611 |
| Token penalaran | 1,529,395 | 46,321 |
| Waktu respons (rata-rata) | 73.07s | 21.06s |
| Waktu respons (maks) | 280.52s | 100.41s |
| Waktu respons (total) | 1096.07s | 315.95s |
Skor vs Total Biaya
Waktu respons (rata-rata)
Skor Rata-rata vs Waktu respons (rata-rata)
Rincian Kategori
| Trik anti-AI | Skor | Konsistensi | Tingkat lulus per percobaan | Tes tidak stabil | Tes benar | Waktu respons (rata-rata) | Token output | Token penalaran |
|---|---|---|---|---|---|---|---|---|
| Google: Gemini 3.1 Flash Lite Preview | 10.0 | 10.0 | 100.0% | 0 | 43.87s | 144 | 193,077 | |
| OpenAI: GPT-5.4 | 10.0 | 10.0 | 100.0% | 0 | 5.02s | 216 | 1,466 |
| Gabungan | Skor | Konsistensi | Tingkat lulus per percobaan | Tes tidak stabil | Tes benar | Waktu respons (rata-rata) | Token output | Token penalaran |
|---|---|---|---|---|---|---|---|---|
| Google: Gemini 3.1 Flash Lite Preview | 10.0 | 10.0 | 100.0% | 0 | 280.52s | 335 | 380,440 | |
| OpenAI: GPT-5.4 | 10.0 | 10.0 | 100.0% | 0 | 20.57s | 301 | 3,543 |
| Parsing dan ekstraksi data | Skor | Konsistensi | Tingkat lulus per percobaan | Tes tidak stabil | Tes benar | Waktu respons (rata-rata) | Token output | Token penalaran |
|---|---|---|---|---|---|---|---|---|
| Google: Gemini 3.1 Flash Lite Preview | 9.9 | 10.0 | 100.0% | 0 | 7.16s | 279 | 6,186 | |
| OpenAI: GPT-5.4 | 9.9 | 10.0 | 100.0% | 0 | 5.32s | 234 | 804 |
| Spesifik domain | Skor | Konsistensi | Tingkat lulus per percobaan | Tes tidak stabil | Tes benar | Waktu respons (rata-rata) | Token output | Token penalaran |
|---|---|---|---|---|---|---|---|---|
| Google: Gemini 3.1 Flash Lite Preview | 4.0 | 10.0 | 33.3% | 0 | 127.58s | 18 | 566,202 | |
| OpenAI: GPT-5.4 | 4.0 | 7.2 | 44.4% | 1 | 74.27s | 61 | 34,748 |
| Kepatuhan instruksi | Skor | Konsistensi | Tingkat lulus per percobaan | Tes tidak stabil | Tes benar | Waktu respons (rata-rata) | Token output | Token penalaran |
|---|---|---|---|---|---|---|---|---|
| Google: Gemini 3.1 Flash Lite Preview | 9.0 | 6.9 | 66.7% | 1 | 70.07s | 69 | 190,053 | |
| OpenAI: GPT-5.4 | 10.0 | 10.0 | 100.0% | 0 | 3.11s | 93 | 897 |
| Puzzle Solving | Skor | Konsistensi | Tingkat lulus per percobaan | Tes tidak stabil | Tes benar | Waktu respons (rata-rata) | Token output | Token penalaran |
|---|---|---|---|---|---|---|---|---|
| Google: Gemini 3.1 Flash Lite Preview | 7.0 | 10.0 | 66.7% | 0 | 46.33s | 87 | 190,953 | |
| OpenAI: GPT-5.4 | 7.0 | 7.2 | 88.9% | 1 | 9.13s | 442 | 3,832 |
| Pemanggilan alat | Skor | Konsistensi | Tingkat lulus per percobaan | Tes tidak stabil | Tes benar | Waktu respons (rata-rata) | Token output | Token penalaran |
|---|---|---|---|---|---|---|---|---|
| Google: Gemini 3.1 Flash Lite Preview | 10.0 | 10.0 | 100.0% | 0 | 7.73s | 234 | 2,484 | |
| OpenAI: GPT-5.4 | 10.0 | 10.0 | 100.0% | 0 | 13.28s | 264 | 1,031 |
Perbandingan Cepat
Ganti Pasangan Perbandingan
Gemini 3 Flash PreviewlowvsGPT-5.4mediumGemini 3.1 Flash Lite PreviewhighvsQwen3.5-122B-A10BmediumGemini 3.1 Flash Lite PreviewhighvsGPT-5.2 ChatnoneGemini 3.1 Flash Lite PreviewhighvsQwen3.5-27BmediumDeepSeek V3.2mediumvsGemini 3.1 Flash Lite PreviewhighClaude Sonnet 4.6mediumvsGemini 3.1 Flash Lite PreviewhighGemini 3.1 Flash Lite PreviewlowvsGPT-5.4mediumGemini 3.1 Flash Lite PreviewhighvsGPT-5.3-CodexmediumGemini 3.1 Flash Lite PreviewhighvsStep 3.5 FlashmediumTersedia gratisGemini 3.1 Flash Lite PreviewhighvsMiMo-V2-FlashmediumGemini 3.1 Flash Lite PreviewhighvsGLM 5mediumGemini 3.1 Flash Lite PreviewhighvsGPT-5.3 Chatnone