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 4.5Mistral Medium 3.1
Claude Haiku 4.5

Anthropic

Mistral Medium 3.1

Mistral

Pricing
Input $/1M$1.00$0.40
Output $/1M$5.00$2.00
Blended $/1M (3:1)$2.00$0.80
Cache read $/1M$0.10$0.04
Cache write $/1M$1.25
Quality indices
Intelligence Index29.614.7
Coding Index43.920.5
Math Index38.3
Agentic Index16.46.2
Speed & latency
Output speed (tok/s)84
Time to first token0.51s
Time to first answer token0.51s
Benchmarks (Artificial Analysis)
MMLU-Pro68.3%
GPQA Diamond58.8%
Humanity's Last Exam4.4%
LiveCodeBench40.6%
SciCode33.8%
AIME 202538.3%
IFBench39.8%
AA-LCR (Long Context Reasoning)19.7%
Terminal-Bench Hard10.6%
Terminal-Bench 2.113.9%
τ²-Bench (Telecom)40.6%
τ³-Bench Banking8.5%
Design Arena (Elo)
Website1,1491,160
UI components1,1421,140
Data viz1,1601,184
SVG1,0721,039
3D1,1311,140
Game dev1,1551,132
Code categories1,1481,154
ASCII art1,1821,040
Specs
Context window200K131K
Max output tokens64K
Input modalitiestext, image, filetext, image, file
TokenizerClaudeMistral