Benchmarks / Knowledge

MedScribe

Can models support doctors with their administrative work?

97models
–Open Index weight
2026-10-13last updated
Vals AIpublisher

MedScribe

Higher is better · best effort per model

Chart settings
97 of 97 selected

full table below

#ModelEffortScoreCost / taskHarness
1Opus 5.5max91.4%$1.15–
2Fable 5.1max91.3%$0.964–
3Sonnet 5.5max91.1%$0.509–
4Opus 5max91.0%$0.236–
5Muse Spark 1.2xhigh90.1%$0.038–
6Haiku 5.5max89.5%$0.044–
7Grok 4.7xhigh89.4%$0.075–
8Wonder Fullmax89.3%$0.192–
9GLM 5.3 Flashmax88.9%$0.0022–
10Muse Spark 1.1xhigh88.9%$0.035–
11GLM 5.3max88.8%$0.057–
12Fable 5max88.5%$0.583–
13MiMo V2.6 Prohigh88.3%$0.0099–
14GPT-5.1high88.1%$0.097–
15Kimi K3max88.0%$0.110–
16GPT-6 Astramax87.9%$0.582–
17Gemini 4high87.4%$0.147–
18MiniMax M3max87.3%$0.014–
19Grok 4.5high86.9%$0.033–
20GPT-5.5xhigh86.9%$0.143–
21Opus 4.6max86.7%$0.115–
22Grok 4.6high86.5%$0.037–
23GPT-6.1 Solmax86.5%$0.101–
24Muse Spark–85.9%$0.0077–
25Opus 4.8max85.8%$0.259–
26DeepSeek V4.1 Flashhigh85.5%$0.015–
27Inklingmax85.4%$0.166–
28Opus 4.5high85.3%$0.410–
29MiMo V2.6 Flashhigh85.3%$0.0022–
30GPT-5.6 Solmax85.2%$0.277–
31Haiku 4.5xhigh85.2%$0.042–
32Qwen 3.8 Maxmax84.9%$0.090–
33Sonnet 4.5high84.5%$0.055–
34Gemini 3.8 Flashhigh84.5%$0.025–
35GPT-5.6 Lunamax84.4%$0.023–
36GPT-5.2xhigh84.4%$0.115–
37Inkling Smallxhigh84.1%$0.019–
38Gemini 3.7 Flashhigh83.9%$0.059–
39Qwen 3.8 27Bxhigh83.8%$0.047–
40MiMo V2.5 Proxhigh83.7%$0.0060–
41GPT-6 Lunamax83.7%$0.0096–
42GPT-5high83.7%$0.102–
43Hy4–83.6%$0.053–
44GLM 5.2max83.5%$0.045–
45Gemini 2.5 Flashmedium83.0%$0.015–
46Opus 4.7max83.0%$0.178–
47GPT-5.6 Terraxhigh82.9%$0.063–
48GPT-6 Solmax82.0%$0.083–
49Grok 4 Fastnone81.6%$0.0025–
50Ling 3.0 Flash–80.9%$0.0014–
51MiniMax M2.1–80.8%$0.0051–
52GPT-5 Minihigh80.6%$0.033–
53Mistral Large 4high80.4%$0.047–
54DeepSeek V4 Flash 0731high80.4%$0.014–
55DeepSeek V4 Pro 0813max80.2%$0.041–
56MiniMax M2.7high79.9%$0.0051–
57Gemini 3.6 Flashhigh79.7%$0.073–
58Qwen 3.7 Maxmax79.4%$0.069–
59Grok 4.1 Fasthigh78.7%$0.0024–
60Grok 4high78.2%$0.064–
61Kimi K2.6max78.1%$0.056–
62GPT-5.4xhigh77.5%$0.639–
63Qwen 3high77.1%$0.020–
64GPT-5.4 Nanohigh77.1%$0.0018–
65Qwen 3.6 Plusmax77.0%$0.029–
66o3high76.7%$0.040–
67Gemini 3.5 Flashhigh76.6%$0.166–
68Kimi K2.5max76.4%$0.025–
69Gemini 3.1 Prohigh76.1%$0.098–
70Sonnet 5max76.1%$0.434–
71Gemini 2.5 Flash Litenone75.8%$0.0013–
72Ling 3.0 Flash Fin–75.6%$0.0016–
73DeepSeek V4max75.1%$0.054–
74Grok 4.3xhigh74.4%$0.015–
75Opus 4.1high73.9%$0.263–
76Gemini 2.5 Prohigh73.6%$0.046–
77GPT-5 Nanohigh72.9%$0.0070–
78Qwen 3 Maxmax72.7%$0.085–
79Sonnet 4high72.4%$0.039–
80GLM 5.1max72.3%$0.024–
81MiMo V2.5xhigh72.2%$0.0014–
82Gemini 3 Prohigh72.0%$0.061–
83Gemini 3.5 Flash-Litehigh70.9%$0.020–
84Qwen 3.5 Flashmax70.6%$0.0044–
85Gemini 3 Flashhigh69.9%$0.014–
86o4 minihigh69.1%$0.041–
87GLM 4.7none68.6%$0.019–
88Mistral Medium 3.5high67.7%$0.158–
89Laguna M.1–65.9%$0.0022–
90Gemini 3.1 Flash-Litehigh63.9%$0.0022–
91Grok 4.20xhigh63.4%$0.031–
92Laguna XS.2–61.4%$0.0011–
93Command Anone55.7%$0.140–
94Mercury 2.5high55.1%$0.0045–
95Llama 4 Maverickmedium54.2%$0.0025–
96Llama 4 Scoutmedium50.6%$0.0017–
97Nemotron 3.5 Lightning–4.3%$0.0062–