Leaderboard / poolside
Laguna XS.2
poolside · Laguna · released 2026-04-28 · data checked 2026-11-01
Laguna XS.2 is a 33B-parameter Mixture-of-Experts model (3B active per token) from poolside (United States), released 2026-04-28 under the Apache-2.0 license. It doesn't have enough reported benchmarks for an index score yet. It needs about 21.6 GB of memory at 4-bit with an 8K-token context.
Agentic coding model. Mean pass@1 over 4-7 runs.
Small LM Index
—
Needs 2 core benchmarks (has 1)
Parameters
33B
3B active · MoE
Price per 1M tokens
—
No hosted API on OpenRouter; self-host
Context window
262K
tokens
Benchmarks
Each bar shows this model's score. The tick marks the best medium model on that benchmark. Sources: 2 reported by poolside. How we label sources.
Specifications
- Lab
- poolside (US · United States)
- Architecture
- Mixture-of-Experts
- Total parameters
- 33B
- Active parameters
- 3B
- Size class
- Medium
- Context window
- 262K tokens
- Input modalities
- text
- Reasoning
- Instruct (no thinking)
- License
- Apache-2.0
- Openness
- Open weights
- Memory to run
- ~21.6 GB at 4-bit, 8K context (more sizes)
- Hugging Face
- 930.1K downloads (30 days) · 320 likes
Compare with
- Laguna XS.2 vs Nanbeige4.3 33B (33.1B, index 77.8)
- Laguna XS.2 vs Qwen3 32B (32.8B, index 52.4)
- Laguna XS.2 vs DeepSeek-R1-Distill-Qwen-32B (32.8B, index 50.9)
Sources
Something wrong here? Report it.
Run it locally
No GGUF build from a well-known publisher yet.
MLX (Mac)
pip install -U mlx-lm
mlx_lm.chat --model mlx-community/Laguna-XS.2-4bitvLLM (GPU server)
vllm serve poolside/Laguna-XS.2Full-precision weights, for a server GPU.
Memory needed: weights, context cache and runtime buffers.
| Quantization | 4K context | 32K context | 256K context |
|---|---|---|---|
| 16-bit F16 | 67 GB | 69 GB | 78 GB |
| 8-bit Q8_0 | 36 GB | 38 GB | 47 GB |
| 4-bit Q4_K_M | 21 GB | 23 GB | 32 GB |