Updated
Claude Sonnet 5.5 vs GPT-6 Astra
Claude Sonnet 5.5 had 25% fewer counted bugs per line than GPT-6 Astra (0.96 vs 1.28).
| Claude Sonnet 5.5 | GPT-6 Astra | |
|---|---|---|
| Bug rate | ||
| Relative bug rate | 0.96 | 1.28 |
| Rank | #2 of 8 | #4 of 8 |
| By bug type | ||
| Logic bugs | 0.74 | 1.14 |
| Testing and docs(statistically tied) | 1.29 | 1.80 |
| By language | ||
| TypeScript | 0.54 | 1.76 |
| Python(statistically tied) | 1.70 | 1.49 |
| Details | ||
| Provider | Anthropic | OpenAI |
| Released | Sep 28, 2026 | Sep 4, 2026 |
| List price | $2 input / $10 output per 1M tokens | $10 input / $50 output per 1M tokens |
| Share of AI-written code, week of Sep 28 | 2.4% | 9.7% |
1.00 = average · lower is better · = statistically tied
More comparisons
Questions and answers
- Which has fewer bugs, Claude Sonnet 5.5 or GPT-6 Astra?
- Claude Sonnet 5.5 had 25% fewer counted bugs per line than GPT-6 Astra (0.96 vs 1.28). Claude Sonnet 5.5 ranks #2 of 8 and GPT-6 Astra #4.
- Which is cheaper, Claude Sonnet 5.5 or GPT-6 Astra?
- Claude Sonnet 5.5 lists at $2 input / $10 output per 1M tokens; GPT-6 Astra lists at $10 input / $50 output per 1M tokens. Claude Sonnet 5.5 has the lower list price.
- Does it depend on the programming language?
- In TypeScript, Claude Sonnet 5.5 had the lower rate (0.54 vs 1.76); in Python, they were statistically tied (1.70 and 1.49).
- Does it depend on the kind of bug?
- For logic bugs, Claude Sonnet 5.5 had the lower rate (0.74 vs 1.14); for testing and documentation issues, they were statistically tied (1.29 and 1.80).