Inference
EN
Vals AI

Code Migration

Can language models reimplement real-world programs in another language?

ModelsModels with a result.77
Top resultBest result on this benchmark.69.8%Claude Sonnet 5.5
Top-3 spreadPoints from first to third.2.1 pts
UpdatedDate the source changed the results.29 Oct 2026

Result and cost

77 of 77 results

Results

77 results
#ModelResultThe score from the source.CostUS dollars to run one task.TimeTime to run one task.
01Claude Sonnet 5.569.8% ±4.3$46.403.3 h
02Gemini 4 Argon68.2% ±4.3$57.822.2 h
03GPT-6 Astra67.7% ±4.2$44.3659 min
04Claude Opus 5.566.6% ±4.3$69.483.3 h
05GPT-6.1 Sol65.1% ±4.4$6.511.6 h
06Muse Wonder62.6% ±4.4$69.542.9 h
07Muse Wonder Full61.0% ±4.4$38.413.9 h
08Claude Opus 557.5% ±4.4$60.512.9 h
09GPT-6 Sol57.2% ±4.2$15.6886 min
10Claude Fable 555.1% ±4.6$112.101.8 h
11Claude Fable 5.154.6% ±4.8$70.974.3 h
12GPT-5.6 Sol52.9% ±4.4$24.5459 min
13GPT-5.6 Terra47.8% ±4.3$8.1349 min
14Hy4 preview47.4% ±4.3$3.412.2 h
14Muse Spark 1.3 Max47.4% ±4.3$14.2184 min
16Claude Opus 4.847.3% ±4.2$30.5175 min
17DeepSeek V4.1 Flash 101745.6% ±4.3$0.9451.8 h
18GPT-5.545.2% ±4.2$6.4432 min
19Grok 4.744.8% ±4.2$36.5569 min
20Grok 4.644.6% ±4.5$14.581.5 h
20Claude Haiku 5.544.6% ±4.2$6.2383 min
22GPT-5.6 Luna44.5% ±4.2$1.8859 min
23Claude Sonnet 544.4% ±4.2$35.312.0 h
24GLM-5.344.2% ±4.3$24.913.8 h
25Claude Opus 4.743.9% ±4.2$36.3972 min
26MiMo-V2.6-Pro43.0% ±4.3$0.6922.8 h
27GPT-6 Luna42.6% ±4.4$0.60150 min
28DeepSeek V4 Pro 081341.5% ±4.3$18.583.5 h
29MiMo-V2.6-Flash40.9% ±4.4$0.4943.2 h
30Claude Sonnet 4.639.9% ±4.1$38.511.7 h
31Step 5 Preview39.0% ±4.5$7.613.6 h
32DeepSeek V4 Flash 073138.6% ±4.4$3.612.1 h
33GLM-5.237.9% ±4.1$12.641.9 h
34Grok 4.536.6% ±4.1$3.3723 min
35Gemini 3.8 Flash36.5% ±4.2$18.492.6 h
36GPT-5.435.0% ±4.1$7.7636 min
37Gemini 3.7 Flash34.8% ±4.2$21.462.6 h
38Ember-132.5% ±4.6$51.114.1 h
39Muse Spark 1.131.1% ±4.1$4.5639 min
40Gemini 3.6 Flash30.9% ±4.1$8.931.5 h
41Mistral Large 430.6% ±4.2$24.524.8 h
42Muse Spark 1.230.0% ±4.0$3.7831 min
43Kimi K2.627.8% ±4.1$5.1190 min
44Muse Spark 1.327.6% ±4.1$4.6268 min
45Gemini 3.5 Flash26.7% ±4.1$5.7319 min
46DeepSeek V4 Pro 042326.2% ±4.0$1.0729 min
47GLM-5.125.8% ±4.1$5.7058 min
48Kimi K2.7 Code25.4% ±4.2$5.041.7 h
49Qwen3.8 Max24.0% ±4.3$10.544.5 h
50MiMo-V2.5-Pro21.6% ±4.1$0.23849 min
51Muse Spark Open21.2% ±3.5$9.041.9 h
52GLM-5.3-Flash20.5% ±4.2$6.263.4 h
53MiniMax M319.9% ±3.9$7.0774 min
54Gemini 3.1 Pro Preview17.3% ±3.9$1.6112 min
55Kimi K316.1% ±4.1$13.874.3 h
56GPT-5.4 nano14.5% ±4.0$0.41627 min
57MiMo-V2.514.2% ±3.7$0.10338 min
57Qwen3.8-27B14.2% ±4.2$21.684.8 h
59Inkling-Small13.7% ±3.8$0.59721 min
60Qwen3.7 Max13.1% ±2.9$1.9326 min
61GPT-5.4 mini12.9% ±3.6$1.9529 min
61Qwen3.7 Plus12.9% ±2.9$0.41936 min
63Inkling11.8% ±3.4$3.0130 min
64Qwen3.6 Plus11.1% ±1.3$4.7830 min
65MiniMax M2.78.7% ±1.8$0.58821 min
66Claude Haiku 4.5 Thinking7.6% ±1.1$0.98210 min
67Kimi K2.5 Thinking7.0% ±1.3$0.92764 min
68Grok 4.36.8% ±1.2$0.4173 min
69Gemini 3 Flash Preview6.4% ±1.1$0.345 min
70Gemini 3.5 Flash-Lite6.1% ±1.7$0.52522 min
71Mistral Medium 3.55.1% ±1.4$18.623.6 h
72Gemini 3.1 Flash Lite Preview4.6% ±1.1$0.0532 min
73Mercury 2.54.5% ±1.0$0.62716 min
74Nemotron Lightning 3p5 30B A3B3.0% ±1.0$0.26743 min
75Ling 3.0 Flash Af Rc31.8% ±0.9$0.10330 min
76Grok 4.20 (Reasoning)0.3% ±0.1$0.2896 min
77Ling 3.0 Flash0.0%$0.065.0 h