Multi-agent software engineering coordination on SWE-bench distinct-shard topology
34.2CF (median)AutoGen
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| AutoGenN=16, Backbone=GPT-4o-mini, n=82, Topology=distinct-shard (SCR = 0)2026.05 | 34.2 | 90.2 | |
| AutoGenN=8, Backbone=GPT-4o-mini, n=88, Topology=distinct-shard (SCR = 0)2026.05 | 30.6 | 88.6 | |
| AutoGenN=4, Backbone=GPT-4o-mini, n=83, Topology=distinct-shard (SCR = 0)2026.05 | 28.1 | 88 | |
| CrewAIN=8, Backbone=GPT-4o-mini, n=102, Topology=distinct-shard (SCR = 0)2026.05 | 9.81 | 41.2 | |
| CrewAIN=16, Backbone=GPT-4o-mini, n=101, Topology=distinct-shard (SCR = 0)2026.05 | 9.31 | 41.6 | |
| CrewAIN=4, Backbone=GPT-4o-mini, n=103, Topology=distinct-shard (SCR = 0)2026.05 | 9.18 | 39.8 | |
| LangGraphN=16, Backbone=GPT-4o-mini, n=132, Topology=distinct-shard (SCR = 0)2026.05 | 7.88 | 82.6 | |
| LangGraphN=8, Backbone=GPT-4o-mini, n=133, Topology=distinct-shard (SCR = 0)2026.05 | 6.79 | 75.9 | |
| LangGraphN=4, Backbone=GPT-4o-mini, n=137, Topology=distinct-shard (SCR = 0)2026.05 | 6.18 | 70.1 | |
| S-BUSN=4, Backbone=GPT-4o-mini, n=138, Topology=distinct-shard (SCR = 0)2026.05 | 0.135 | 73.9 | |
| S-BUSN=8, Backbone=GPT-4o-mini, n=133, Topology=distinct-shard (SCR = 0)2026.05 | 0.133 | 71.4 | |
| S-BUSN=16, Backbone=GPT-4o-mini, n=132, Topology=distinct-shard (SCR = 0)2026.05 | 0.126 | 70.5 |