Add a benchmark

Benchmarks / Coding

Code Migration

Can language models reimplement real-world programs in another language? Vals AI's overall score for Code Migration.

76models
–Open Index weight
2026-10-12last updated
Vals AIpublisher

Code Migration

Higher is better · best effort per model

Chart settings
76 of 76 selected

full table below

#ModelEffortScoreCost / taskHarness
1Sonnet 5.5max69.8%$75.83–
2Gemini 4high68.2%$57.82–
3GPT-6 Astramax67.7%$44.36–
4Opus 5.5max66.6%$112.97–
5GPT-6.1 Solmax65.1%$6.51–
6Wondermax62.6%$69.54–
7Wonder Fullmax61.0%$38.41–
8Opus 5max57.5%$60.51–
9GPT-6 Solmax57.2%$15.68–
10Fable 5max55.1%$112.10–
11Fable 5.1max54.6%$70.97–
12GPT-5.6 Solmax52.9%$24.54–
13GPT-5.6 Terramax47.8%$8.13–
14Hy4–47.4%$3.41–
15Muse Spark 1.3max47.4%$14.21–
16Opus 4.8max47.3%$30.51–
17DeepSeek V4.1 Flashhigh45.6%$0.945–
18GPT-5.5xhigh45.2%$6.44–
19Grok 4.7xhigh44.8%$36.55–
20Grok 4.6high44.6%$14.58–
21Haiku 5.5max44.6%$6.23–
22GPT-5.6 Lunamax44.5%$1.88–
23Sonnet 5max44.4%$35.31–
24GLM 5.3max44.2%$24.91–
25Opus 4.7max43.9%$36.39–
26MiMo V2.6 Prohigh43.0%$0.692–
27GPT-6 Lunamax42.6%$0.601–
28DeepSeek V4 Pro 0813max41.5%$18.58–
29MiMo V2.6 Flashhigh40.9%$0.494–
30Sonnet 4.6max39.9%$38.51–
31Step 5 Preview–39.0%$7.61–
32DeepSeek V4 Flash 0731high38.6%$3.61–
33GLM 5.2max37.9%$12.64–
34Grok 4.5high36.6%$3.37–
35Gemini 3.8 Flashhigh36.5%$18.49–
36GPT-5.4xhigh35.0%$7.76–
37Gemini 3.7 Flashhigh34.8%$21.46–
38Ember-1–32.5%$51.11–
39Muse Spark 1.1xhigh31.1%$4.56–
40Gemini 3.6 Flashhigh30.9%$8.93–
41Mistral Large 4high30.6%$47.89–
42Muse Spark 1.2xhigh30.0%$3.78–
43Kimi K2.6max27.8%$5.11–
44Gemini 3.5 Flashhigh26.7%$5.73–
45DeepSeek V4max26.2%$1.07–
46GLM 5.1max25.8%$5.70–
47Kimi K2.7 Codemax25.4%$5.04–
48Qwen 3.8 Maxmax24.0%$10.54–
49MiMo V2.5 Proxhigh21.6%$0.238–
50GLM 5.3 Flashmax20.5%$3.13–
51MiniMax M3max19.9%$7.07–
52Gemini 3.1 Prohigh17.3%$1.61–
53Kimi K3max16.1%$13.87–
54GPT-5.4 Nanohigh14.5%$0.416–
55MiMo V2.5xhigh14.2%$0.103–
56Qwen 3.8 27Bxhigh14.2%$21.68–
57Inkling Smallxhigh13.7%$0.597–
58Qwen 3.7 Maxmax13.1%$1.93–
59GPT-5.4 Minixhigh12.9%$1.95–
60Qwen 3.7 Plushigh12.9%$0.419–
61Inklingmax11.8%$3.01–
62Qwen 3.6 Plusmax11.1%$4.78–
63MiniMax M2.7high8.7%$0.588–
64Haiku 4.5xhigh7.6%$0.982–
65Kimi K2.5max7.0%$0.927–
66Grok 4.3high6.8%$0.417–
67Gemini 3 Flashhigh6.4%$0.340–
68Gemini 3.5 Flash-Litehigh6.1%$0.525–
69Mistral Medium 3.5high5.1%$18.62–
70Nemotron 3 Ultra 550B A55Bmax4.9%––
71Gemini 3.1 Flash-Litehigh4.6%$0.053–
72Mercury 2.5high4.5%$0.627–
73Nemotron 3.5 Lightning–3.0%$0.267–
74Ling 3.0 Flash Fin–1.8%$0.103–
75Grok 4.20xhigh0.3%$0.289–
76Ling 3.0 Flash–0.0%$0.060–