Compare Models
Select models (max 5)
Claude Haiku 5.5DeepSeek V4.1 Flash 1017
 |  |  |
|---|---|---|
| Vals IndexGDP-weighted benchmark | 54.44%±1.41 | 56.58%±1.27 |
| Legal Research BenchAgentic US legal research | 43.27%±3.44 | 45.19%±3.46 |
| Finance Agent (v2)Core financial analyst tasks | 54.14%±2.13 | 56.96%±2.05 |
| Tax Agent BenchAgentic US corporate tax research | 31.92%±3.35 | 44.35%±3.57 |
| MedCodeMedical billing code support | 51.54%±2.17 | N/A |
| Terminal-Bench-ScienceExpert-authored scientific research workflows | 10.00%±3.61 | N/A |
| Code MigrationRewriting programs in new languages | 44.57%±4.23 | 44.21%±4.46 |
| Terminal-Bench 4.0Frontier-difficulty terminal tasks | 36.36%±2.31 | 31.82%±2.62 |
| Vibe Code Bench v1.1Building web apps from scratch | 90.44%±1.73 | 87.73%±2.75 |
Benchmarks
Vals Index *
Claude Haiku 5.5
0.00%± 1.41
(50/50)DeepSeek V4.1 Flash 1017
0.00%± 1.27
(50/50)Legal Research Bench *
Claude Haiku 5.5
0.00%± 3.44
(80/80)DeepSeek V4.1 Flash 1017
0.00%± 3.46
(80/80)Finance Agent (v2) *
Claude Haiku 5.5
0.00%± 2.13
(80/80)DeepSeek V4.1 Flash 1017
0.00%± 2.05
(80/80)Tax Agent Bench *
Claude Haiku 5.5
0.00%± 3.35
(71/71)DeepSeek V4.1 Flash 1017
0.00%± 3.57
(71/71)MedCode *
Claude Haiku 5.5
0.00%± 2.17
(107/107)DeepSeek V4.1 Flash 1017
N/ATerminal-Bench-Science
Claude Haiku 5.5
0.00%± 3.61
(41/41)DeepSeek V4.1 Flash 1017
N/ACode Migration *
Claude Haiku 5.5
0.00%± 4.23
(79/79)DeepSeek V4.1 Flash 1017
0.00%± 4.46
(79/79)Terminal-Bench 4.0
Claude Haiku 5.5
0.00%± 2.31
(50/50)DeepSeek V4.1 Flash 1017
0.00%± 2.62
(50/50)Vibe Code Bench v1.1 *
Claude Haiku 5.5
0.00%± 1.73
(114/114)DeepSeek V4.1 Flash 1017
0.00%± 2.75
(114/114)Overall performance
Performance on the Vals Index, a GDP-weighted aggregation of tasks across finance, coding, and law
•2026-10-18
Comparison by Industry
Model performance on different sections of the economy.
Legal
22.26%
vs.27.18%
Finance
48.36%
vs.54.37%
Healthcare
70.54%
vs.N/A
Math
89.00%
vs.N/A
Science
38.50%
vs.N/A
Education
45.78%
vs.N/A
Coding
40.79%
vs.54.59%
Cyber
40.60%
vs.N/A
Social Mobility
63.19%
vs.N/A
Cost Analysis
Per task and per token model pricing.
Vals Index · USD
$3.00
$0.637
Cost / TestVals Index
$3.00
vs.$0.637
Input Cost/ 1M Tokens
$0.10
vs.$0.30
Input Cache Write/ 1M Tokens
$0.125
vs.N/A
Input Cache Read/ 1M Tokens
$0.01
vs.$0.006
Output Cost/ 1M Tokens
$0.50
vs.$1.20
Model Metadata
Basic information about each model.
| | |
|---|---|---|
| Model provider | Anthropic | DeepSeek |
| Latency | 2843.64s | 2193.12s |
| Cost (In/Out) | $0.1 / $0.5 | $0.3 / $1.2 |
| Context Window | 1M | 1M |
| Max Output Token | 128k | 384k |
| Input Modality |