Navigasi
AI BENCHY
Advertise here

Anthropic: Claude Opus 4.8 vs DeepSeek: DeepSeek V4 Flash

Claude Opus 4.8 (low) unggul dalam skor rata-rata dengan 7.8 vs 7.7. DeepSeek V4 Flash (high) memiliki biaya benchmark lebih rendah di $0.042 vs $2.077. Claude Opus 4.8 (low) lebih cepat di 12.74s vs 49.75s, dengan tingkat keberhasilan 80.3% vs 72.7%.

Model yang direkomendasikanDeepSeek V4 Flash (high)Its score stays close to the best score here (7.7 vs 7.8), while costing about 49.5x less than Claude Opus 4.8 (low).

Benchmark dihasilkan dari suite pengujian AI BENCHY pada: 2026-07-17

Metrik Claude Opus 4.8 Claude Opus 4.8 low Rilis: 2026-05-28 DeepSeek V4 Flash DeepSeek V4 Flash high Rilis: 2026-04-24
Skor 7.8 7.7
Peringkat #39 #43
Keandalan 10.0 10.0
Konsistensi 8.9 8.2
Tes benar
Tingkat lulus per percobaan 80.3% 72.7%
Tes tidak stabil 3 5
Total Run 66 66
Biaya per hasil 12.979 0.402
Total Biaya $2.077 $0.042
Harga input $5.000 / 1M $0.098 / 1M
Harga output $25.000 / 1M $0.196 / 1M
Total token input 156,525 108,392
Token output 43,141 14,478
Token penalaran 8,617 153,687
Waktu respons (rata-rata) 12.74s 49.75s
Waktu respons (maks) 127.97s 218.13s
Waktu respons (total) 280.29s 1094.41s

Hamster playing table tennis

Prompt: Create a detailed SVG illustration of a hamster playing table tennis.

#39 Claude Opus 4.8

low
Biaya
$0.031
Waktu
14.1s
Token
1,345 tok

#43 DeepSeek V4 Flash

high
Biaya
$0.003
Waktu
93.1s
Token
7,926 tok

Model teratas berdasarkan skor

Skor vs Total Biaya

Waktu respons (rata-rata)

Skor vs Waktu respons (rata-rata)

Total token output

Skor vs Total token output

Rincian Kategori

Pemrograman Skor Konsistensi Tingkat lulus per percobaan Tes tidak stabil Tes benar Waktu respons (rata-rata) Token input Token output Token penalaran
Claude Opus 4.8 6.6 4.6 77.8% 2 7.58s 10,590 3,637 809
DeepSeek V4 Flash 7.8 10.0 66.7% 0 50.60s 7,279 395 34,862

Perbandingan Cepat

Ganti Pasangan Perbandingan

Claude Opus 4.6mediumvsDeepSeek V4 FlashhighClaude Opus 4.8lowvsGLM 5mediumDeepSeek V4 FlashhighvsGLM 5mediumDeepSeek V4 FlashhighvsMiniMax M3mediumClaude Opus 4.8lowvsGPT-5.6 LunahighDeepSeek V4 FlashhighvsGrok Build 0.1mediumClaude Opus 4.8lowvsGPT-5.6 TerramediumClaude Opus 4.8lowvsGLM 5.2mediumClaude Opus 4.8lowvsQwen3.6 PlusmediumClaude Opus 4.8lowvsDeepSeek V4 ProhighDeepSeek V4 FlashhighvsGPT-5.6 LunamediumClaude Opus 4.8lowvsQwen3.7 Plusmedium