Laguna S 2.1
PoolsideBalanced
ThinkingTool Use
About this model
Laguna S 2.1 is the latest coding agent model from [Poolside](<https://poolside.ai/>). Laguna S 2.1 is a 118B total parameter model with 8B active parameters, scoring 70.2% on Terminal-Bench 2.1 and...
Performance Tier
Balanced
Laguna S 2.1 is a balanced model from Poolside : strong performance at a reasonable price.
Strong cost-performance ratio. Reliable for most professional use cases without premium pricing.
Pricing
This model is included in Elosia plansEco
Minimal cost. Ideal for very high volume or simple tasks.
| Type | per 1M tokens |
|---|---|
| Input (prompt) | $0.090 |
| Output (completion) | $0.180 |
| Cache read | $0.0090 |
Capabilities
Context Length1.0M
Max Output Tokens131K
TokenizerOther
Inputtext
Outputtext
Release DateJuly 21, 2026
Benchmarks
General Intelligence
MMLU
Not reported
Mathematics
MATH-500
Not reported
Programming
HumanEval
Not reported
SWE-bench Multilingual
78.5%
Agentic
SWE-bench Pro
59.4%
Terminal-Bench 2.1
70.2%
Where does Laguna S 2.1 stand?
Compare its performance index against every other model.
Recommended Use Cases
Coding
Strengths
- Most capable agentic coding model in its weight class: SWE-bench Multilingual 78.5%, SWE-bench Pro 59.4%
- Outperforms models over 10× its size on agentic coding despite a compact 118B / 8B-active Mixture-of-Experts
- Strong terminal and tool-use workflows (Terminal-Bench 2.1 70.2%)
- Open weights (OpenMDW-1.1) with a 1M-token context and FP8 / NVFP4 / INT4 / GGUF quantized builds
Limitations
- Code specialist: no published general-knowledge or math benchmarks
- Benchmarks are self-reported by Poolside (no independent leaderboard)
Frequently asked questions
Resources
This model may use your data for training
