Compare Models

Select models (max 7)
Qwen 3.8 MaxDeepSeek V4 Flash 0731
Benchmarks

Legal Research Bench *

Qwen 3.8 Max
0.00%± 3.47
(98/98)
DeepSeek V4 Flash 0731
0.00%± 3.19
(98/98)

Tax Agent Bench *

Qwen 3.8 Max
0.00%± 3.37
(89/89)
DeepSeek V4 Flash 0731
0.00%± 3.16
(89/89)

MedCode *

Qwen 3.8 Max
0.00%± 2.03
(107/107)
DeepSeek V4 Flash 0731
0.00%± 2.15
(107/107)

Terminal-Bench-Science

Qwen 3.8 Max
0.00%± 4.03
(47/47)
DeepSeek V4 Flash 0731
0.00%± 2.44
(47/47)

Terminal-Bench 4.0

Qwen 3.8 Max
0.00%± 3.94
(68/68)
DeepSeek V4 Flash 0731
0.00%± 3.54
(68/68)

Vibe Code Bench v1.1 *

Qwen 3.8 Max
0.00%± 5.36
(132/132)
DeepSeek V4 Flash 0731
0.00%± 3.44
(132/132)

Overall performance

Performance on the Vals Index, a GDP-weighted aggregation of tasks across finance, coding, legal, and tax

2026-11-16

Comparison by Industry

Model performance on different sections of the economy.

Legal
67.11%
54.00%
Finance
59.41%
49.38%
Healthcare
62.81%
60.89%
Math
58.00%
56.00%
Science
13.98%
34.37%
Academic
90.11%
88.05%
Education
81.76%
N/A
Coding
53.24%
50.41%
Cyber
28.57%
72.08%
Social Mobility
73.02%
N/A

Cost Analysis

Per task and per token model pricing.

Vals Index · USD
N/A
N/A
Cost / TestVals Index
N/A
N/A
Input Cost/ 1M Tokens
$2.00
$0.30
Input Cache Write/ 1M Tokens
$2.50
N/A
Input Cache Read/ 1M Tokens
$0.25
$0.006
Output Cost/ 1M Tokens
$6.00
$1.20

Model Metadata

Basic information about each model.

Model provider
AlibabaAlibaba
DeepSeekDeepSeek
LatencyN/AN/A
Cost (In/Out)$2 / $6$0.3 / $1.2
Context Window1M1M
Max Output Token128k384k
Input Modality