Frontier Log

Poolside

Laguna M.1

Released · poolside.ai

Context window

TokensContext window
Default256K
Context history
DateContext window
poolside.ai · 2 · 3 · Corroborated256K

Benchmark results

BenchmarkScoreSource verified
SWE-bench Verifiedpoolside harness on Harbor; pass@1 averaged over 4 runs; some task images patched72.5%Corroborated The same number is printed in 1 other lab document: 1
poolside harness on Harbor; pass@1 averaged over 4 runs; some task images patched · reported later, in Long context update: Laguna XS.2 and M.174.6%Corroborated The same number is printed in 1 other lab document: 1
SWE-bench Propoolside harness on Harbor; pass@1 averaged over 3 runs; some task images patched · The source doesn't say whether this is the public or the commercial set46.9%Corroborated The same number is printed in 1 other lab document: 1
poolside harness on Harbor; pass@1 averaged over 3 runs; some task images patched · reported later, in Long context update: Laguna XS.2 and M.149.2%Corroborated The same number is printed in 1 other lab document: 1
Terminal-Bench 2.0poolside harness on Harbor; pass@1 averaged over 5 runs; 48GB RAM/32 CPUs; some task images patched40.7%Corroborated The same number is printed in 1 other lab document: 1
poolside harness on Harbor; pass@1 averaged over 5 runs; 48GB RAM/32 CPUs; some task images patched · reported later, in Long context update: Laguna XS.2 and M.145.8%Corroborated The same number is printed in 1 other lab document: 1
Sources

Reported by Poolside

Related

All models