Professional work benchmarks
Evaluate legal, financial, healthcare, and other specialist work. Check each task's domain, required expertise, and scoring method when interpreting model performance.
Published
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.2Extra high40%
- 2Muse Spark 1.1Extra high40%
- 3Muse Spark 1.3Extra high20%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark OpenExtra high20%
- 2Muse Spark 1.2Extra high20%
- 3MiMo-V2.6-Flash20%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Gemini 3.7 FlashHigh20%
- 2Muse WonderMax20%
- 3Gemini 3.8 FlashHigh20%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.3Extra high40%
- 2Muse Spark 1.1Extra high20%
- 3Gemini 3.7 FlashHigh0%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.2Extra high60%
- 2Muse Spark 1.3Extra high40%
- 3Gemini 4 ArgonHigh40%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark OpenExtra high40%
- 2Muse Spark 1.3Extra high40%
- 3Claude Fable 5Max20%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1DeepSeek V4 Pro 0813Max20%
- 2Muse Spark 1.1Extra high20%
- 3DeepSeek V4.1 Flash 1017Max20%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.2Extra high94.52%
- 2Gemini 4 ArgonHigh94.04%
- 3Muse Spark 1.1Extra high92.86%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.3Extra high96.38%
- 2Muse Spark 1.2Extra high95.29%
- 3Gemini 4 ArgonHigh94.2%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.2Extra high96.06%
- 2Muse Spark 1.1Extra high94.27%
- 3Gemini 4 ArgonHigh94.27%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.2Extra high97.77%
- 2Gemini 4 ArgonHigh95.91%
- 3Muse Spark 1.1Extra high94.8%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.3Extra high95.5%
- 2Muse Spark 1.2Extra high93.77%
- 3Muse Spark 1.1Extra high92.04%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.2Extra high96.86%
- 2Gemini 4 ArgonHigh95.69%
- 3Muse Spark 1.3Extra high94.51%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Gemini 4 ArgonHigh97.26%
- 2Muse Spark 1.1Extra high96.92%
- 3Muse Spark 1.2Extra high96.23%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.1Extra high96.15%
- 2Gemini 4 ArgonHigh96.15%
- 3Muse Spark 1.2Extra high95.83%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Gemini 4 ArgonHigh97.53%
- 2Muse WonderMax97.53%
- 3Claude Fable 5.1Max96.82%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.3Extra high95.56%
- 2Muse WonderMax95.56%
- 3Muse Spark 1.2Extra high95.22%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.2Extra high89.51%
- 2Muse Spark 1.3Extra high89.51%
- 3Gemini 4 ArgonHigh88.76%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.2Extra high98.68%
- 2Claude Opus 4.8Max97.03%
- 3DeepSeek V4 Pro 0813Max96.7%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.3Extra high95.4%
- 2Gemini 4 ArgonHigh95.34%
- 3Muse Spark 1.2Extra high93.68%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Gemini 4 ArgonHigh93%
- 2Muse Spark 1.1Extra high92.13%
- 3Muse WonderMax91.25%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.2Extra high95.38%
- 2Muse WonderMax94.06%
- 3Muse Spark 1.3Extra high93.73%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse WonderMax94.12%
- 2Muse Spark 1.2Extra high94.12%
- 3Muse Spark 1.1Extra high93.21%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.2Extra high94.14%
- 2Claude Fable 5Max93.41%
- 3Gemini 4 ArgonHigh93.04%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Gemini 4 ArgonHigh96.47%
- 2Muse Spark 1.2Extra high95.76%
- 3Muse Spark 1.1Extra high95.41%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.3Extra high94.55%
- 2Claude Fable 5.1Max94.05%
- 3Gemini 4 ArgonHigh93.25%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.2Extra high95.13%
- 2Muse Spark 1.3Extra high94.38%
- 3Muse Spark 1.1Extra high93.63%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.2Extra high96.5%
- 2Muse WonderMax96.21%
- 3Gemini 4 ArgonHigh95.92%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse WonderMax90.48%
- 2Muse Spark 1.3Extra high90.16%
- 3Muse Spark 1.2Extra high88.89%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Gemini 4 ArgonHigh93.31%
- 2Muse Spark 1.2Extra high92.98%
- 3Muse Spark 1.1Extra high91.3%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse WonderMax93.89%
- 2Muse Spark 1.2Extra high93.51%
- 3Gemini 4 ArgonHigh93.51%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.3Extra high94.28%
- 2Gemini 4 ArgonHigh93.94%
- 3Muse Spark 1.2Extra high93.27%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark OpenExtra high40%
- 2Muse Spark 1.3Extra high40%
- 3Muse Spark 1.2Extra high40%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Gemini 3.8 FlashHigh40%
- 2Muse Spark 1.3Extra high40%
- 3Muse Spark 1.2Extra high40%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark OpenExtra high0%
- 2Muse Spark 1.1Extra high0%
- 3Claude Opus 4.7Max0%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Step 5 Preview40%
- 2Muse Spark 1.2Extra high40%
- 3DeepSeek V4 Pro 0813Max40%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.3Extra high0%
- 2Gemini 4 ArgonHigh0%
- 3Laguna Xs 20%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.1Extra high40%
- 2Gemini 4 ArgonHigh40%
- 3MiMo-V2.6-Flash20%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.2Extra high60%
- 2Muse Spark 1.3Extra high40%
- 3DeepSeek V4 Flash 0731High40%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.2Extra high40%
- 2Muse Spark 1.1Extra high20%
- 3Gemini 4 ArgonHigh20%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Gemini 3.7 FlashHigh20%
- 2Claude Opus 5.5Max0%
- 3Gemini 3.6 FlashHigh0%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Gemini 4 ArgonHigh20%
- 2DeepSeek V4.1 Flash 1017Max20%
- 3Muse Spark OpenExtra high20%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.3Extra high20%
- 2Gemini 4 ArgonHigh20%
- 3Muse Spark OpenExtra high20%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse WonderMax40%
- 2DeepSeek V4.1 Flash 1017Max20%
- 3MiMo-V2.6-Pro20%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.2Extra high25.42%
- 2Muse Spark 1.3Extra high22.92%
- 3Muse Spark 1.1Extra high20%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.3Extra high20%
- 2Muse WonderMax20%
- 3Claude Opus 4.7Max20%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1MiMo-V2.6-Flash20%
- 2DeepSeek V4 Flash 0731High20%
- 3Muse Spark 1.2Extra high20%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.2Extra high20%
- 2Muse Spark OpenExtra high20%
- 3Gemini 4 ArgonHigh0%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Gemini 4 ArgonHigh40%
- 2Muse Spark 1.3Extra high20%
- 3DeepSeek V4.1 FlashMax20%
Updated:
Models tested: 49
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1
- 1Muse Spark 1.2Extra high0%
- 2Claude Opus 4.7Max0%
- 3Claude Sonnet 5Max0%
Top models · Vals AI
Latest result per model · 1
- 1Gemini 4 ArgonHigh82.98%
- 2Claude Fable 5.1Max82.23%
- 3Claude Sonnet 5.5Max80.04%
Top models · Vals AI
Latest result per model · 1
- 1Claude Opus 5Max87.68%
- 2Claude Fable 5.1Max86.74%
- 3Gemini 3.7 FlashHigh86.48%
Top models · Vals AI
Latest result per model · 1
- 1Claude Opus 5.5Max70.14%
- 2Claude Sonnet 5.5Max70.07%
- 3Claude Opus 5Max70.02%
Top models · Vals AI
Latest result per model · 1
- 1Claude Sonnet 5.5Max69.98%
- 2Claude Fable 5.1Max69.64%
- 3Claude Opus 5.5Max69.01%
Top models · Vals AI
Latest result per model · 1
- 1GPT-6.1 SolMax83.91%
- 2GPT-5.6 SolMax83.87%
- 3Gemini 3.7 FlashHigh83.36%
Top models · Vals AI
Latest result per model · 1
- 1Gemini 4 ArgonHigh73.95%
- 2Claude Fable 5.1Max72.15%
- 3Claude Fable 5Max71.4%
Top models · Vals AI
Latest result per model · 1
- 1Claude Opus 5.5Max77.86%
- 2GPT-6 AstraMax75.44%
- 3Claude Fable 5.1Max74.85%
Top models · Vals AI
Latest result per model · 1
- 1Claude Fable 5.1Max76.67%
- 2Claude Opus 5.5Max75.94%
- 3Claude Sonnet 5.5Max75.71%
Top models · Vals AI
Latest result per model · 1
- 1Claude Fable 5.1Max72.59%
- 2Claude Opus 5Max72.26%
- 3Gemini 4 ArgonHigh70.8%
Top models · Vals AI
Latest result per model · 1
- 1Claude Opus 5.5Max81.39%
- 2Claude Sonnet 5.5Max80.94%
- 3Claude Fable 5.1Max80.8%
Updated:
Models tested: 49
Can models support doctors with their administrative work?
Top models · Vals AI
Latest result per model · 1
- 1Claude Opus 5.5Max91.43%
- 2Claude Fable 5.1Max91.29%
- 3Claude Sonnet 5.5Max91.1%
Top models · Vals AI
Latest result per model · 1
- 1Claude Opus 5Max63.57%
- 2Gemini 3.1 Pro PreviewHigh59.06%
- 3Gemini 4 ArgonHigh58.8%
Updated:
Models tested: 34
Can AI help people navigate SNAP benefits?
Top models · Vals AI
Latest result per model · 1.1
- 1Claude Opus 5Max76.93%
- 2Claude Fable 5.1Max74.9%
- 3Claude Opus 5.5Max70.64%
Top models · Vals AI
Latest result per model · 1
- 1Gemma 4 31BHigh50.79%
- 2Gemini 3.5 Flash-LiteHigh49.21%
- 3DeepSeek V4.1 FlashHigh48.41%
Top models · Vals AI
Latest result per model · 1
- 1Gemini 4 ArgonHigh45.64%
- 2Gemini 2.5 Flash44.05%
- 3Gemini 2.5 Pro43.65%
Top models · Vals AI
Latest result per model · 1
- 1Claude Opus 4.8Max78.25%
- 2Claude Opus 4.7Max77.04%
- 3Claude Opus 5Max76.06%
Top models · Vals AI
Latest result per model · 1
- 1Claude Opus 4.7Max56.1%
- 2Gemma 4 31BHigh55.03%
- 3Claude Opus 4.8Max54.79%
Updated:
Models tested: 19
Evaluating agents on Excel-based financial modeling tasks, graded by the GPT-6 Luna rubric judge with template scoring fixes
Top models · Vals AI
Latest result per model · 1.1
- 1Gemini 4 ArgonHigh82.94%
- 2Claude Fable 5.1Max82%
- 3Claude Sonnet 5.5Max79.41%
Updated:
Models tested: 19
Evaluating agents on Excel-based financial modeling tasks, graded by the GPT-6 Luna rubric judge with template scoring fixes
Top models · Vals AI
Latest result per model · 1.1
- 1Claude Fable 5.1Max88.52%
- 2Gemini 3.8 FlashHigh87.8%
- 3Muse WonderMax87.73%
Updated:
Models tested: 19
Evaluating agents on Excel-based financial modeling tasks, graded by the GPT-6 Luna rubric judge with template scoring fixes
Top models · Vals AI
Latest result per model · 1.1
- 1Claude Sonnet 5.5Max72.97%
- 2Claude Fable 5.1Max71.52%
- 3Claude Opus 5.5Max69.87%
Updated:
Models tested: 19
Evaluating agents on Excel-based financial modeling tasks, graded by the GPT-6 Luna rubric judge with template scoring fixes
Top models · Vals AI
Latest result per model · 1.1
- 1Claude Sonnet 5.5Max68.66%
- 2Claude Opus 5.5Max68.33%
- 3Claude Fable 5.1Max67.99%
Updated:
Models tested: 19
Evaluating agents on Excel-based financial modeling tasks, graded by the GPT-6 Luna rubric judge with template scoring fixes
Top models · Vals AI
Latest result per model · 1.1
- 1GPT-6.1 SolMax83.08%
- 2GPT-6 AstraMax82.9%
- 3Claude Sonnet 5.5Max82.29%
Updated:
Models tested: 19
Evaluating agents on Excel-based financial modeling tasks, graded by the GPT-6 Luna rubric judge with template scoring fixes
Top models · Vals AI
Latest result per model · 1.1
- 1Gemini 4 ArgonHigh73.04%
- 2Claude Fable 5.1Max71.36%
- 3Claude Opus 5.5Max71%
Updated:
Models tested: 19
Evaluating agents on Excel-based financial modeling tasks, graded by the GPT-6 Luna rubric judge with template scoring fixes
Top models · Vals AI
Latest result per model · 1.1
- 1Claude Opus 5.5Max77.74%
- 2GPT-6 AstraMax75.51%
- 3Claude Fable 5.1Max75.18%
Updated:
Models tested: 19
Evaluating agents on Excel-based financial modeling tasks, graded by the GPT-6 Luna rubric judge with template scoring fixes
Top models · Vals AI
Latest result per model · 1.1
- 1Claude Fable 5.1Max76.73%
- 2Claude Sonnet 5.5Max75.9%
- 3Claude Opus 5.5Max75.84%
Updated:
Models tested: 19
Evaluating agents on Excel-based financial modeling tasks, graded by the GPT-6 Luna rubric judge with template scoring fixes
Top models · Vals AI
Latest result per model · 1.1
- 1Claude Fable 5.1Max72.67%
- 2GPT-6 AstraMax70.87%
- 3Claude Sonnet 5.5Max70.82%
Updated:
Models tested: 19
Evaluating agents on Excel-based financial modeling tasks, graded by the GPT-6 Luna rubric judge with template scoring fixes
Top models · Vals AI
Latest result per model · 1.1
- 1Claude Opus 5.5Max81.39%
- 2Claude Sonnet 5.5Max80.94%
- 3Claude Fable 5.1Max80.8%
Top models · Vals AI
Latest result per model · 1
- 1Gemini 4 ArgonHigh84.88%
- 2Gemini 3.7 FlashHigh83.64%
- 3Claude Opus 5.5Max83.64%
Updated:
Models tested: 14
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1.1
- 1Gemini 4 ArgonHigh60%
- 2DeepSeek V4.1 Flash 1017Max30%
- 3Muse WonderMax20%
Updated:
Models tested: 14
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1.1
- 1MiMo-V2.6-Flash30%
- 2MiMo-V2.6-Pro20%
- 3DeepSeek V4 Pro 0813Max20%
Updated:
Models tested: 14
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1.1
- 1Gemini 4 ArgonHigh40%
- 2Muse WonderMax30%
- 3GPT-6.1 SolMax10%
Updated:
Models tested: 14
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1.1
- 1MiMo-V2.6-Pro20%
- 2Muse WonderMax20%
- 3DeepSeek V4 Pro 0813Max20%
Updated:
Models tested: 14
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1.1
- 1Gemini 4 ArgonHigh40%
- 2Muse WonderMax30%
- 3Gemini 3.8 FlashHigh30%
Updated:
Models tested: 14
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1.1
- 1Gemini 4 ArgonHigh60%
- 2Muse WonderMax30%
- 3MiMo-V2.6-Pro20%
Updated:
Models tested: 14
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1.1
- 1Muse WonderMax40%
- 2Gemini 4 ArgonHigh30%
- 3Gemini 3.8 FlashHigh30%
Updated:
Models tested: 14
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1.1
- 1Gemini 4 ArgonHigh96.3%
- 2Muse WonderMax96.01%
- 3MiMo-V2.6-Pro92.88%
Updated:
Models tested: 14
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1.1
- 1Gemini 4 ArgonHigh97.65%
- 2Muse WonderMax95.65%
- 3MiMo-V2.6-Pro93.66%
Updated:
Models tested: 14
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1.1
- 1MiMo-V2.6-Flash96.95%
- 2DeepSeek V4 Pro 0813Max96.24%
- 3Gemini 4 ArgonHigh96.06%
Updated:
Models tested: 14
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1.1
- 1Gemini 4 ArgonHigh97.39%
- 2Muse WonderMax96.46%
- 3MiMo-V2.6-Flash94.03%
Updated:
Models tested: 14
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1.1
- 1Gemini 4 ArgonHigh95.16%
- 2Muse WonderMax94.46%
- 3DeepSeek V4.1 Flash 1017Max93.43%
Updated:
Models tested: 14
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1.1
- 1Gemini 4 ArgonHigh96.67%
- 2Gemini 3.8 FlashHigh96.47%
- 3Muse WonderMax96.28%
Updated:
Models tested: 14
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1.1
- 1Gemini 4 ArgonHigh97.95%
- 2Muse WonderMax97.6%
- 3Gemini 3.8 FlashHigh93.84%
Updated:
Models tested: 14
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1.1
- 1Muse WonderMax97.91%
- 2Gemini 4 ArgonHigh97.27%
- 3Gemini 3.8 FlashHigh96.95%
Updated:
Models tested: 14
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1.1
- 1Gemini 4 ArgonHigh98.41%
- 2MiMo-V2.6-Pro97.7%
- 3Muse WonderMax97.7%
Updated:
Models tested: 14
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1.1
- 1Muse WonderMax97.44%
- 2Gemini 4 ArgonHigh96.42%
- 3Claude Sonnet 5.5Max95.56%
Updated:
Models tested: 14
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1.1
- 1Muse WonderMax93.07%
- 2Gemini 4 ArgonHigh92.88%
- 3DeepSeek V4.1 Flash 1017Max90.64%
Updated:
Models tested: 14
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1.1
- 1Muse WonderMax98.52%
- 2Gemini 4 ArgonHigh98.52%
- 3DeepSeek V4.1 Flash 1017Max98.02%
Updated:
Models tested: 14
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1.1
- 1Gemini 4 ArgonHigh97.31%
- 2Muse WonderMax94.8%
- 3DeepSeek V4.1 Flash 1017Max93.01%
Updated:
Models tested: 14
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1.1
- 1Muse WonderMax95.04%
- 2Gemini 4 ArgonHigh93%
- 3Gemini 3.8 FlashHigh90.53%
Updated:
Models tested: 14
Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools
Top models · Vals AI
Latest result per model · 1.1
- 1Gemini 4 ArgonHigh97.82%
- 2MiMo-V2.6-Pro95.97%
- 3DeepSeek V4.1 Flash 1017Max95.97%
Methodology and provenance
Benchize collects and structures public benchmark reports. Scores, costs, timings, ranks and uncertainty remain attributed to their original sources. Different versions, configurations and measurement units are separate results. Missing fields stay unavailable.