Laguna XS.2

+ contribute

Laguna XS.2 is an open-weights language model from Poolside, released April 2026. Its published size is 33B parameters (3B active per token) and a 262k-token context window. Licence: apache-2.0.

Catalogue identity notes

Model card (poolside/Laguna-XS.2 @a397bde): '33B total with 3B activated per token', Apache 2.0; not the later Laguna-XS-2.1. Epoch model_metadata.csv date 2026-04-28, 'Open weights (unrestricted)'. Arena entry 'laguna-xs.2' (poolside, Apache 2.0). Official model card documents vLLM 0.21.0+ serving; BF16 weight-file sizes verified at the pinned revision, not total runtime memory.

64.6±1.5#150 of 719 LLMsfrom 8 benchmark results across 3 sources

$ how it compares

The LLMs ranked either side of Laguna XS.2, out of 828 with a score, then more from Poolside.

#	model	score	where	RAM	released
170	Solar Pro 4 Upstage	65±2	api $0.53/M		2026-08
171	MiniMax-M2 MiniMax	65±1	api $0.53/M +local	~75GB	2025-10
172	Laguna XS.2 Poolside	65±2	local	~21GB	2026-04
173	Qwen3 235B-A22B Instruct 2507 Alibaba	65±1	api $0.20/M +local	~76GB	2025-07
174	Mistral Medium 3.5 (128B) Mistral	65±1	api $3.00/M +local	~42GB	2026-04
	more from Poolside
130	Laguna M.1 Poolside	67±2	api price unknown +local	~135GB	2026-04
156	Laguna XS 2.1 (33B-A3B) Poolside	66±2	local	~21GB	2026-06

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasks█···█······█ #3 of 3 · 65% of its score · 4 tests

test	result	vs the others in its run	difficulty	counts	why
SWE-bench Verified lab +1	69.9% ✎	▁▁▁▁▁▁▂▄█▆▁▁ beats 75/131	hard · avg 27%	▇▇▇▇▇▇▇▇▇ 24%	sharp · at its level · small run · several runs
Terminal-Bench 2.0 lab +1	35.7% ✎	▂▂▅▅████▆▂▃▃ beats 8/36	very hard · avg 11%	▇▇▇▇▇▇▇▇ 20%	at its level · small run · several runs
SWE-bench Multilingual lab +1	57.7% ✎	▁··▁▃▂▅▇█▃▃▁ beats 18/60	hard · avg 28%	▇▇▇▇▇▇▇ 18%	at its level · small run · several runs
SWE-bench Pro lab +1	46.3% ✎	▅▃▃·▃·▅▅▅█▃▃ beats 1/15	hard · avg 30%	▇ 3.5%	blunt · at its level · small run

coding▁▁▁▃▆██▇▆▄▃▁ #128 of 458 · 79% of its score · 4 tests

Beats Qwen3.5 Plus, Gemini 2.5 Pro, Gemini 3.1 Flash-Lite here, though they rank higher overall. Trails Llama 3.1 Nemotron Ultra 253B v1, Nemotron 3.5 Lightning 30B-A3B, Gemma 4 12B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Arena WebDev (Code Arena)	1301.9 ✎	▁▂▄▃▅█▄▇▆▄▁▁ beats 21/119	avg 981	▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 34%	sharp
SWE-bench Verified lab +1	69.9% ✎	▁▁▁▁▁▁▂▄█▆▁▁ beats 75/131	hard · avg 27%	▇▇▇▇▇▇▇▇▇ 24%	sharp · at its level · small run · several runs
SWE-bench Multilingual lab +1	57.7% ✎	▁··▁▃▂▅▇█▃▃▁ beats 18/60	hard · avg 28%	▇▇▇▇▇▇▇ 18%	at its level · small run · several runs
SWE-bench Pro lab +1	46.3% ✎	▅▃▃·▃·▅▅▅█▃▃ beats 1/15	hard · avg 30%	▇ 3.5%	blunt · at its level · small run

$ where to run it

llama.cpp · recommended

Install llama.cpp · Homebrew and winget installs name the commands llama-cli and llama-server instead.
macOS / Linux
curl -LsSf https://llama.app/install.sh | sh
Windows
irm https://llama.app/install.ps1 | iex
other ways to install ↗
Run Laguna XS.2 · needs ~20.3 GB · from huggingface.co/Lucebox
chat in the terminal
llama cli -hf Lucebox/Laguna-XS.2-GGUF:Q4_K_M
Or start a local server and open the address it prints (http://localhost:9931)
llama serve -hf Lucebox/Laguna-XS.2-GGUF:Q4_K_M

LM Studio

version

Install LM Studio · Open the app once before using the lms command.
download it (macOS / Windows / Linux) ↗
Run Laguna XS.2 · needs ~20.3 GB · from huggingface.co/Lucebox
open in LM Studio ↗
Or download it in the terminal
lms get https://huggingface.co/Lucebox/Laguna-XS.2-GGUF@Q4_K_M -y
Then run lms ls to see its name, and lms chat followed by that name

Ollama

version

Install Ollama
download it (macOS / Windows) ↗
Linux
curl -fsSL https://ollama.com/install.sh | sh
Run Laguna XS.2 · needs ~23 GB · from ollama.com
chat in the terminal
ollama run laguna-xs.2:latest

MLX (Mac)

version

Install MLX (Mac) · Apple silicon Macs only, text only.
with uv
uv tool install mlx-lm
or
pip install mlx-lm
Run Laguna XS.2 · needs ~18.8 GB, Mac only · from huggingface.co/mlx-community
chat in the terminal
mlx_lm.chat --model mlx-community/Laguna-XS.2-4bit
Or start an OpenAI-compatible server at http://localhost:8080/v1
mlx_lm.server --model mlx-community/Laguna-XS.2-4bit

Other ways to run it

Run it as published · needs 66.89 GB, NVIDIA only · from huggingface.co/poolside (official)
vllm serve poolside/Laguna-XS.2 --tool-call-parser poolside_v1 --reasoning-parser poolside_v1 --enable-auto-tool-choice

Official weights and documentation: huggingface.co/poolside ↗

$ how the score was worked out · Poolside vendor-reported benc… 36%, LMArena leaderboard dataset 34%, Poolside — Laguna M.1/XS.2 Te… 30%

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

source	results	share of the score	licence
Poolside vendor-reported benchmark scores for its own Laguna models (Hugging Face model cards) · lab-reported	4	▇▇▇▇▇▇▇▇▇▇▇ 36%	Factual vendor-reported score (not licensed content)
LMArena leaderboard dataset	1	▇▇▇▇▇▇▇▇▇▇ 34%	CC BY 4.0
Poolside — Laguna M.1/XS.2 Technical Report · lab-reported	3	▇▇▇▇▇▇▇▇▇ 30%	CC BY-SA 4.0
prior · pull toward the middle		▇ 1%
weighted average		= 64.6

not counted (1)

poolside/Laguna-XS.2 model card

SWE-bench Pro [correct]

46.3%

the only model in this source's run of the test, so there is nothing to compare it with there; another run of the test counts

$ improve this page

Update model information Add a place to run it Add a test result Propose a benchmark Correct a benchmark score Edit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated 2026-10-11 02:20 UTC · sources and licences · json