Expired50% confidenceFactTime unknown
On the SWE-bench software engineering task benchmark, the task completion rates of Claude 3.5 Sonnet and GPT-4o far exceed those of open-source models.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/2/2026
First Seen
Valid until: 9/30/2026(expired)
Sources
Related Claims
UnverifiedModels such as Claude 3.5 Sonnet and GPT-4o are able to resolve a considerable proportion of real GitHub issues on SWE-bench (Software Engineering Benchmark)85% similarUnverifiedFor routine code completion tasks, GPT-4o-mini or Claude Sonnet is sufficient and there is no need to consume top-tier model credits.79% similarUnverifiedThe DeepSeek series scores comparably to GPT-4o and Claude Sonnet on some programming tasks in public benchmarks such as HumanEval and SWE-bench.77% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/54191API
curl https://kongchang.com/api/v1/knowledge/claims/54191MCP
get_claim(id=54191)