SWE-bench Lite leaderboard
- models measured
- 6
- lead over #2
- +10.0 pts
- median
- 35.2%
Leader on SWE-bench Lite · read October 21
# SWE-bench Lite, read October 21
leaders = {
"Qwen3 Coder 30B A3B Instruct": 49.7%, "GPT-4": 39.7%, "GPT-4o": 38.0%,
}
On October 21, 2026, 6 of our catalog's models had a SWE-bench Lite result, and Qwen3 Coder 30B A3B Instruct led with 49.7%.
A 300-issue subset of SWE-bench chosen to be cheaper to run: a real GitHub issue from an open source Python project, and a patch that has to make the project tests pass. The number is the share of issues resolved.
Role in the index
A coding board we show beside the Vibe Coding Index, not inside it: the index keeps the four boards it was defined on, so a score does not move every time a board is added. Read the methodology
| # | Model | Result | Evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|
| 1 | Qwen3 Coder 30B A3B InstructQwen | 49.7% | 60.4 | $0.28 | |
| 2 | GPT-4OpenAI | 39.7% | 22.4 | $60.00 | |
| 3 | GPT-4oOpenAI | 38.0% | 38.8 | $10.00 | |
| 4 | o3 MiniOpenAI | 32.3% | 52.1 | $4.40 | |
| 5 | DeepSeek V3DeepSeek | 30.7% | 70.2 | $1.0287 | |
| 6 | GPT-4o (2024-08-06)OpenAI | 21.7% | 23.1 | $10.00 |
Only models in our catalog are listed, and each appears once with its best published run. 38 rows were read from the source on October 21, 2026, and rows that match no catalog model are left out.