Laguna S 2.1

Poolside

Laguna S 2.1 is Poolside's most capable Laguna 2.1 model: an open-weight Mixture-of-Experts coding model with 118B total and 8B activated parameters (256 routed experts plus one shared). It targets long-horizon and agentic coding with interleaved thinking and a 1,048,576-token context window in thinking and no-thinking modes. Text-to-text only, available as open weights and via Poolside's API and OpenRouter.

Key Specifications

Parameters
118.0B
Context
1.0M
Release Date
July 21, 2026
Average Score
57.4%

Timeline

Key dates in the model's history
Announcement
July 21, 2026
Last Update
August 31, 2026

Technical Specifications

Parameters
118.0B
Training Tokens
-
Knowledge Cutoff
-
Family
-
Capabilities
MultimodalZeroEval

Pricing & Availability

Input (per 1M tokens)
$0.10
Output (per 1M tokens)
$0.20
Max Input Tokens
1.0M
Max Output Tokens
-
Supported Features
Function CallingStructured OutputCode ExecutionWeb SearchBatch InferenceFine-tuning

Benchmark Results

Model performance metrics across various tests and benchmarks

Other Tests

Specialized benchmarks
DeepSWE 1.1
Harbor fork matching DeepSWE Pier harness; thinking on; mean pass@1 over 3 attempts; as of 21 July 2026Self-reported
40.4%
SWE Atlas - Codebase QnA
Scale AI SWE Atlas Codebase QnA via Harbor; Opus 4.5 judge; mean pass@1 over 3 attempts; as of 21 July 2026Self-reported
46.2%
SWE-bench Multilingual
Harbor + Poolside agent harness; thinking on; mean pass@1 over 4 attempts; as of 21 July 2026Self-reported
78.5%
SWE-Bench Pro
Harbor + Poolside agent harness; thinking on; mean pass@1 over 4 attempts; public set; as of 21 July 2026Self-reported
59.4%
Terminal-Bench 2.1
Harbor + Poolside agent harness; thinking on; mean pass@1 over 4 attempts; as of 21 July 2026Self-reported
70.2%
Toolathlon
Toolathlon Verified; Harbor + Poolside agent harness; thinking on; mean pass@1 over 3 attempts; as of 21 July 2026Self-reported
49.7%

License & Metadata

License
openmdw
Announcement Date
July 21, 2026
Last Updated
August 31, 2026

Compare Laguna S 2.1

All comparisons

Similar Models

All Models

Recommendations are based on similarity of characteristics: developer organization, multimodality, parameter size, and benchmark performance. Choose a model to compare or go to the full catalog to browse all available AI models.