Model Comparison & Benchmarks
Capability Radar Matrix
Normalized multi-benchmark capability profiles.
Top for Coding
Claude 3.5 Sonnet
Score of 92.0% on HumanEval. Unmatched understanding of complex multi-file TypeScript and Python architectures.
Top for Vision & Speed
OpenAI GPT-4o
Fastest Time-to-First-Token (28ms) with native cross-modal audio and video understanding.
Best Context Window
Google Gemini 1.5 Pro
Supports up to 2,000,000 context tokens with near 100% needle-in-a-haystack recall.
Empirical Benchmark Scores
| Model | Provider | MMLU Score | HumanEval | Input / 1M | Output / 1M | Context Window |
|---|---|---|---|---|---|---|
| Claude 3.5 Sonnet | Anthropic | 88.7% | 92.0% | $3.00 | $15.00 | 200,000 |
| GPT-4o | OpenAI | 88.9% | 90.2% | $2.50 | $10.00 | 128,000 |
| Gemini 1.5 Pro | 85.9% | 84.1% | $3.50 | $10.50 | 2,000,000 |