Model radar
Compare models across five dimensions simultaneously. Select up to 8 models — overlapping polygons reveal trade-offs at a glance.
5 of 8 max selected
Claude Fable 5
Claude Opus 4.7
Claude Opus 4.8
GPT-5.4
GPT-5.5
Axes explained
- Intelligence
- Artificial Analysis Intelligence Index — composite benchmark across MMLU, GPQA, MATH, HumanEval and others. Raw AA 0–100 score.
- Coding
- Artificial Analysis Coding Index — HumanEval, SWE-bench, and related coding benchmarks. Raw AA 0–100 score.
- Agentic
- Artificial Analysis Agentic Index — tool-use, multi-step, and agent benchmark performance. Raw AA 0–100 score.
- Speed
- Output tokens per second (Artificial Analysis median). Percentile rank within this model set — top percentile = fastest.
- Affordability
- Inverted price percentile: cheaper models score higher. Blended price = input × 0.3 + output × 0.7. Top percentile = cheapest.
Intelligence, Coding, and Agentic use raw AA absolute scores (0–100). Speed and Affordability are percentile-ranked within this dataset so all axes share a 0–100 scale.