Laguna XS.2: benchmark results
Poolside's open agentic-coding MoE (33B total, 3B active) for single-GPU local use. Provider: Poolside. Released 2026-04-28. Access: Open.
Aggregate Rating 1551 ± 24, rank #613 of 1656 rated models, from 48 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| SvelteBench | 95.6 | Average pass@1 (%) | 77 |
| Tinybird AI SQL Benchmark - Success Rate | 100 | Questions answered with a valid query within 3 retries (%) | 71.7 |
| Tinybird AI SQL Benchmark - Exactness | 51.56 | Result exactness vs human reference queries (0-100) | 69.5 |
| Wolfram LLM Benchmarking Project | 47.8 | Correct Functionality (%) | 58.9 |
| SpeechMap Compliance | 59.5 | % Requests Completed | 44.5 |
| Vals AI LiveCodeBench | 67.8 | Accuracy (%) | 33.8 |
| CorpFin v2 | 56.33 | Score (%) | 33.6 |
| Vals AI CorpFin v2 | 56.33 | Accuracy (%) | 32.3 |
| Vals AI Terminal-Bench 2.0 | 28.09 | Accuracy (%) | 28.8 |
| Tinybird AI SQL Benchmark - First-Attempt Success Rate | 86 | Questions answered with a valid query on the first attempt ( | 25.8 |
| Vals AI ProgramBench | 0 | Raw Pass Rate (%) | 25.4 |
| LegalBench | 71.03 | Score (%) | 20.3 |
Interactive version: theaggregate.ai/model?slug=laguna-xs-2 · How It Works · Data refreshed daily, snapshot 2026-10-20.