Compare Models
Select models (max 7)
DeepSeek V4Grok 4.5
 |  |  |
|---|---|---|
| Legal Research BenchAgentic US legal research | 23.08%±2.93 | 37.98%±3.37 |
| Tax Agent BenchAgentic US corporate tax research | 25.17%±3.06 | 24.86%±3.12 |
| MedCodeMedical billing code support | 40.45%±2.12 | 43.29%±2.31 |
| Terminal-Bench 4.0Frontier-difficulty terminal tasks | 11.11%±1.34 | 8.59%±1.34 |
| Vibe Code Bench v1.1Building web apps from scratch | 49.93%±4.77 | 69.00%±4.48 |
Benchmarks
Legal Research Bench *
DeepSeek V4
0.00%± 2.93
(84/84)Grok 4.5
0.00%± 3.37
(84/84)Tax Agent Bench *
DeepSeek V4
0.00%± 3.06
(75/75)Grok 4.5
0.00%± 3.12
(75/75)MedCode *
DeepSeek V4
0.00%± 2.12
(107/107)Grok 4.5
0.00%± 2.31
(107/107)Terminal-Bench 4.0
DeepSeek V4
0.00%± 1.34
(54/54)Grok 4.5
0.00%± 1.34
(54/54)Vibe Code Bench v1.1 *
DeepSeek V4
0.00%± 4.77
(118/118)Grok 4.5
0.00%± 4.48
(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
51.70%
vs.61.97%
Finance
48.62%
vs.52.78%
Healthcare
57.80%
vs.65.09%
Math
16.00%
vs.31.00%
Academic
88.32%
vs.81.31%
Education
N/A
vs.34.97%
Coding
46.20%
vs.51.16%
Cyber
N/A
vs.35.89%
Social Mobility
62.92%
vs.N/A
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
$1.32
vs.$2.00
Input Cache Read/ 1M Tokens
$0.044
vs.$0.30
Output Cost/ 1M Tokens
$3.96
vs.$6.00
Model Metadata
Basic information about each model.
| | |
|---|---|---|
| Model provider | DeepSeek | SpaceXAI |
| Latency | N/A | N/A |
| Cost (In/Out) | $1.32 / $3.96 | $2 / $6 |
| Context Window | 1M | 500k |
| Max Output Token | 384k | 500k |
| Input Modality |