AiCostCompare

Compare models

Pick up to 4 models for a side-by-side spec sheet. Hover any dotted metric name for what the test measures. Best value in each row is highlighted.

Claude Haiku 4.5DeepSeek V4 Pro 0423
Claude Haiku 4.5

Anthropic

DeepSeek V4 Pro 0423

DeepSeek

Pricing
Input $/1M$1.00$0.422
Output $/1M$5.00$0.845
Blended $/1M (3:1)$2.00$0.528
Cache read $/1M$0.10$0.035
Cache write $/1M$1.25
Quality indices
Intelligence Index16.936
Coding Index43.968.8
Math Index
Agentic Index826.3
Speed & latency
Output speed (tok/s)85
Time to first token1.09s
Time to first answer token24.59s
Benchmarks (Artificial Analysis)
GPQA Diamond92.8%
Humanity's Last Exam41.0%
SciCode51.0%
AA-LCR (Long Context Reasoning)80.3%
Terminal-Bench 2.178.7%
τ³-Bench Banking39.6%
Design Arena (Elo)
Website1,1351,247
Web apps1,000
Full-stack948
UI components1,1151,238
Data viz1,1391,218
SVG1,0491,164
3D1,1011,283
Game dev1,1201,252
Godot game dev1,059
Code categories1,1301,256
ASCII art1,1591,167
Specs
Context window200K1.05M
Max output tokens64K384K
Input modalitiestext, image, filetext
TokenizerClaudeDeepSeek