Side-by-side comparison of StarCoder 2 15B (BigCode) and Claude 3 Opus (Anthropic) — benchmarks, pricing, and capabilities.
| StarCoder 2 15B BigCode | Claude 3 Opus Anthropic | |
|---|---|---|
| Category | Code Models | LLMs |
| Specifications | ||
| Context Window | 16K | 200K |
| Pricing (per 1M tokens) | ||
| Input Cost | — | $15.00 |
| Output Cost | — | $75.00 |
| Performance | ||
| Overall Score | 72.0 | 90.5 |
| ARC-Challenge | — | 90.7 |
| BigBench Hard | — | 84.5 |
| Chatbot Arena ELO | — | 1178.0 |
| DROP | — | 85.5 |
| GSM8K | — | 73.7 |
| HumanEval | 44.2 | 84.9 |
| MATH | — | 60.1 |
| MBPP | 64.5 | — |
| MMLU | — | 86.8 |
| TruthfulQA | — | 69.0 |
| WinoGrande | — | 84.0 |
| Community | ||
| User Rating | ★ 4.0 | ★ 4.6 |
| Reviews | 180 | 720 |