Compare Models
Select models (max 7)
Mistral Large 4Step 5 Preview
 |  |  |
|---|---|---|
| Legal Research BenchAgentic US legal research | 31.73%±3.23 | 40.38%±3.41 |
| Finance Agent v2.1Core financial analyst tasks | 61.38%±2.02 | 58.90%±2.05 |
| Tax Agent BenchAgentic US corporate tax research | 30.39%±3.31 | 27.27%±3.11 |
| MedCodeMedical billing code support | 40.69%±2.17 | N/A |
| Terminal-Bench-ScienceExpert-authored scientific research workflows | N/A | 2.86%±2.01 |
| Code Migration v1.1Rewriting programs in new languages | N/A | 47.00%±3.43 |
| Terminal-Bench 4.0Frontier-difficulty terminal tasks | 27.78%±1.82 | 31.82%±2.62 |
| Vibe Code Bench v1.1Building web apps from scratch | 78.40%±3.55 | 70.40%±4.27 |
Benchmarks
Legal Research Bench *
Mistral Large 4
0.00%± 3.23
(84/84)Step 5 Preview
0.00%± 3.41
(84/84)Finance Agent v2.1 *
Mistral Large 4
0.00%± 2.02
(30/30)Step 5 Preview
0.00%± 2.05
(30/30)Tax Agent Bench *
Mistral Large 4
0.00%± 3.31
(75/75)Step 5 Preview
0.00%± 3.11
(75/75)MedCode *
Mistral Large 4
0.00%± 2.17
(107/107)Step 5 Preview
N/ATerminal-Bench-Science
Mistral Large 4
N/AStep 5 Preview
0.00%± 2.01
(42/42)Code Migration v1.1 *
Mistral Large 4
N/AStep 5 Preview
0.00%± 3.43
(28/28)Terminal-Bench 4.0
Mistral Large 4
0.00%± 1.82
(54/54)Step 5 Preview
0.00%± 2.62
(54/54)Vibe Code Bench v1.1 *
Mistral Large 4
0.00%± 3.55
(118/118)Step 5 Preview
0.00%± 4.27
(118/118)Overall performance
Performance on the Vals Index, a GDP-weighted aggregation of tasks across finance, coding, and law
•2026-10-22
Comparison by Industry
Model performance on different sections of the economy.
Legal
24.62%
vs.40.38%
Finance
53.58%
vs.51.91%
Healthcare
60.56%
vs.N/A
Math
10.00%
vs.42.00%
Science
39.60%
vs.29.58%
Coding
39.77%
vs.45.62%
Cyber
N/A
vs.0.38%
Social Mobility
N/A
vs.58.12%
Cost Analysis
Per task and per token model pricing.
Vals Index · USD
N/A
N/A
Cost / TestVals Index
N/A
vs.N/A
Input Cost/ 1M Tokens
$1.36
vs.$1.00
Input Cache Read/ 1M Tokens
$0.14
vs.$0.05
Output Cost/ 1M Tokens
$4.18
vs.$2.70
Model Metadata
Basic information about each model.
| | |
|---|---|---|
| Model provider | Mistral AI | Stepfun |
| Latency | N/A | N/A |
| Cost (In/Out) | $1.36 / $4.18 | $1 / $2.7 |
| Context Window | 512k | 1M |
| Max Output Token | 256k | 1M |
| Input Modality |