Laguna S
by Poolside · Model · Language & code models
Poolside's 118B MoE coding model, 8B active, 1M context, thinking modes; 33B XS variant also released with open weights
AccessOpen weightsNeeds~235 GBDownload235 GBSize118B parametersLicenseOpenMDW-1.1
Evidence
- Jul 26Via Hermes Agent API, found and fixed an unstated seaport-app berth-draft bug, but overthought and hallucinated two extra bugs
- Jul 26Self-reported DeepSWE 40%, slightly above GLM 5.2 which is nearly seven times larger
- Jul 26In the reviewer's tests the XS variant is well behind Qwen 3.6 35B and 27B, the best medium local models
- Jul 22Laguna S 2.1: 118B MoE, 8B active per token, 1M context, native interleaved reasoning, built for long-horizon agentic coding
- Jul 22Model card chart shows Terminal-Bench 70.2 vs 64 for DeepSeek V4 Pro Max; reviewer says he doesn't trust the benchmark