The Agent RaceJuly 23, 2026via The Decoder
Poolside's Laguna S 2.1 is a small open-weight coding model that punches well above its size
Why it matters
Poolside demonstrates that efficient training (agentic reasoning, self-correction) can outperform scale in coding tasks. This challenges the 'bigger = better' narrative and has immediate implications for inference cost economics and open-weight model viability.
Key signals
- Laguna S 2.1 is third coding model released in three months
- Small open-weight model beats larger rivals on benchmarks
- Trained with self-checking, revision loops, and long agentic reasoning chains
- Solved open math problem (unsolved since 1975) for under $0.10
- Focus on efficiency over raw scale
The hook
Small beats big. Poolside's Laguna S 2.1 outperforms much larger coding models—and solved a 51-year-old math problem for under 10 cents.
Poolside has released Laguna S 2.1, its third coding model in three months. Rather than rely on raw scale, the company trained it to keep checking its work, revise failed approaches, and avoid giving up too soon during long agentic sessions. The compact model beats several much larger rivals in benchmarks. Poolside says it also solved a math problem that had been open since 1975 for under 10 cents.