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 Opus 4.6gpt-oss-20b
Claude Opus 4.6

Anthropic

gpt-oss-20b

OpenAI

Pricing
Input $/1M$5.00$0.03
Output $/1M$25.00$0.13
Blended $/1M (3:1)$10.00$0.055
Cache read $/1M$0.50$0.03
Cache write $/1M$6.25
Quality indices
Intelligence Index37.814.9
Coding Index20.7
Math Index89.3
Agentic Index3.1
Speed & latency
Output speed (tok/s)54243
Time to first token1.89s0.39s
Time to first answer token1.89s8.61s
Benchmarks (Artificial Analysis)
MMLU-Pro74.8%
GPQA Diamond84.0%68.8%
Humanity's Last Exam18.6%9.8%
LiveCodeBench77.7%
SciCode45.7%34.4%
AIME 202589.3%
IFBench44.6%65.1%
AA-LCR (Long Context Reasoning)58.3%30.7%
Terminal-Bench Hard48.5%10.6%
Terminal-Bench 2.113.9%
τ²-Bench (Telecom)84.8%60.2%
τ³-Bench Banking7.0%
Design Arena (Elo)
Website1,323880
Web apps1,261
Full-stack1,279
Mobile apps1,280
Android native1,195
UI components1,336
Data viz1,322971
SVG1,274
3D1,335
Game dev1,337
Code categories1,327
ASCII art1,298
Specs
Context window1M131K
Max output tokens128K131K
Input modalitiestext, image, filetext
TokenizerClaudeGPT