unsloth/Laguna-S-2.1-GGUF

Running Unsloth's UD-Q4_K_XL with llama.cpp (PR #25165)

The GGUFs in this repo are Unsloth Dynamic 2.0 quants (imatrix calibrated). Laguna support is not in a tagged llama.cpp release yet, so build llama.cpp from ggml-org/llama.cpp#25165:

git clone https://github.com/ggml-org/llama.cpp
cd llama.cpp
gh pr checkout 25165
# build with CUDA (drop -DGGML_CUDA=ON for a CPU-only build)
cmake -B build -DGGML_CUDA=ON
cmake --build build -j --config Release --target llama-cli llama-server
cd ..

Download the UD-Q4_K_XL shards (~40GB, split into 3 files):

huggingface-cli download unsloth/Laguna-S-2.1-GGUF \
  --include "UD-Q4_K_XL/*" \
  --local-dir Laguna-S-2.1-GGUF

Serve it with llama-server (pass the first shard; the remaining shards load automatically):

./llama.cpp/build/bin/llama-server \
    --model Laguna-S-2.1-GGUF/UD-Q4_K_XL/Laguna-S-2.1-UD-Q4_K_XL-00001-of-00003.gguf \
    --jinja -fa on -ngl 99 --ctx-size 16384 --port 8000

Or run a one-off generation with llama-cli:

./llama.cpp/build/bin/llama-cli \
    --model Laguna-S-2.1-GGUF/UD-Q4_K_XL/Laguna-S-2.1-UD-Q4_K_XL-00001-of-00003.gguf \
    --jinja -ngl 99 -p "Write a Flappy Bird game in Python."

-ngl 99 offloads all layers to GPU; lower it (or drop it) if you run out of VRAM. UD-Q4_K_XL is roughly 40GB, so it fits on a single 48GB+ GPU or splits across several GPUs.

Use on OpenRouter · Use on Vercel AI Gateway · Release blog post

Laguna S 2.1

Laguna S 2.1 is a 118B total parameter Mixture-of-Experts model with 8B activated parameters per token, designed for agentic coding and long-horizon work. It sits between Laguna XS 2.1 (33B-A3B) and Laguna M.1 (225B-A23B) in the Laguna series and shares the family recipe: a token-choice router with softplus gating over 256 routed experts plus one shared expert, grouped-query attention, and interleaved full/sliding-window attention.

Highlights

• Mixed SWA and global attention layout: 48 layers in a 1:3 global-to-SWA ratio (12 global attention layers, 36 sliding-window layers, window 512), with softplus attention gating and per-layer-type rotary scales
• 1M context: 1,048,576-token context window
• Native reasoning support: interleaved thinking between tool calls, with per-request control via enable_thinking
• Speculative decoding: a trained DFlash draft model is available for lower-latency serving
• Quantized variants: FP8, NVFP4, INT4 and GGUF
• OpenMDW-1.1 license: Use and modify the model and associated materials freely for commercial and non-commercial purposes (learn more about OpenMDW)

Model overview

• Number of parameters: 118B total, ~8B activated per token
• Layers: 48 (12 global attention, 36 sliding-window attention)
• Experts: 256 routed (top-10) plus 1 shared expert
• Attention: grouped-query, 8 KV heads, head dim 128; per-head softplus output gating
• Sliding window: 512 tokens
• Context window: 1,048,576 tokens
• Vocabulary: 100,352 tokens (Laguna family tokenizer)
• Modality: text-to-text
• Reasoning: interleaved thinking with preserved thinking

Benchmark results

Model	Size	Terminal-Bench 2.1	SWE-bench Multilingual	SWE-Bench Pro (Public Dataset)	DeepSWE	SWE Atlas (Codebase QnA)	Toolathlon Verified
Laguna S 2.1	118B-A8B	70.2%	78.5%	59.4%	40.4%	46.2%	49.7%
Tencent Hy3	295B-A21B	71.7%	75.8%	57.9%	-	-	-
Inkling	975B-A41B	63.8%	-	54.3%	-	-	45.5%*
Nemotron 3 Ultra	550B-A55B	56.4%	67.7%	-	-	-	34.3%*
DeepSeek-V4-Pro Max	1.6T-A49B	64.0%*	76.2%	55.4%	9.0%*	27.2%*	55.9%*
Kimi K3	2800B-A50B	88.3%	-	-	69%	-	-
Qwen 3.7 Max	-	74.5%*	78.3%	60.6%	-	-	-
Muse Spark 1.1	-	80%	-	61.5%	53.3%	42.2%*	75.6%
Claude Fable 5	-	88%	-	80.3%	70%	-	-

Benchmarks as of 21 July 2026. Laguna S 2.1 in bold; a dash (-) marks a benchmark a model was not evaluated on. Scores marked * are as reported by third parties: Terminal-Bench 2.1 and DeepSWE via Artificial Analysis, SWE Atlas via Scale AI's official leaderboard, and Toolathlon Verified via its official leaderboard. Full evaluation trajectories: trajectories.poolside.ai.

Usage

Laguna S 2.1 uses the same laguna architecture as Laguna XS 2.1, so the same engine integrations apply (vLLM, SGLang, Transformers, TRT-LLM, llama.cpp). At 118B parameters the BF16 checkpoint needs multiple GPUs (roughly 236GB of weights); quantized variants reduce this substantially.

vLLM

vllm serve \
    --model poolside/Laguna-S-2.1 \
    --tensor-parallel-size 4 \
    --tool-call-parser poolside_v1 \
    --reasoning-parser poolside_v1 \
    --enable-auto-tool-choice \
    --served-model-name laguna \
    --default-chat-template-kwargs '{"enable_thinking": true}'

Optional: speculative decoding with DFlash. Pair with the Laguna S 2.1 DFlash draft model by adding --speculative-config '{"model":"poolside/Laguna-S-2.1-DFlash","num_speculative_tokens":7,"method":"dflash"}'.

SGLang

python -m sglang.launch_server \
  --model-path poolside/Laguna-S-2.1 \
  --tp-size 4 \
  --reasoning-parser poolside_v1 \
  --tool-call-parser poolside_v1 \
  --trust-remote-code

TRT-LLM

trtllm-serve poolside/Laguna-S-2.1 --trust-remote-code \
    --tool_parser poolside_v1 --reasoning_parser laguna

Note the flag names differ from vLLM's (--tool_parser, and the reasoning parser is laguna, not poolside_v1).

llama.cpp

GGUF conversions are available at poolside/Laguna-S-2.1-GGUF. Serve with poolside's llama.cpp fork, branch laguna, which carries full Laguna support including DFlash speculative decoding. (Base Laguna support is also in upstream review: ggml-org/llama.cpp#25165.)

git clone --branch laguna https://github.com/poolsideai/llama.cpp
cd llama.cpp && cmake -B build && cmake --build build -j

./build/bin/llama-server -m laguna-s-2.1-Q4_K_M.gguf --jinja --port 8000

# with DFlash speculative decoding:
./build/bin/llama-server -m laguna-s-2.1-Q4_K_M.gguf \
  -md laguna-s-2.1-DFlash-BF16.gguf \
  --spec-type draft-dflash --spec-draft-n-max 15 -fa on --jinja --port 8000

Controlling reasoning

Laguna S 2.1 has native reasoning support and works best with preserved thinking: keep reasoning_content from prior assistant messages in the message history. The model will generally reason before calling tools and between tool calls, and may stop reasoning in follow-up steps if prior thinking blocks are dropped.

Thinking is controlled per request via the chat template:

extra_body={"chat_template_kwargs": {"enable_thinking": False}}

or at the server level with --default-chat-template-kwargs '{"enable_thinking": true}'. For agentic coding use cases we recommend enabling thinking and preserving reasoning in the message history.

License

This model is licensed under the OpenMDW-1.1 License.

Intended and Responsible Use

Laguna S 2.1 is designed for software engineering and agentic coding use cases, and you are responsible for confirming that it is appropriate for your intended application. Laguna S 2.1 is subject to the OpenMDW-1.1 License, and should be used consistently with Poolside's Acceptable Use Policy. We advise against circumventing Laguna S 2.1 safety guardrails without implementing substantially equivalent mitigations appropriate for your use case.

Please report security vulnerabilities or safety concerns to security@poolside.ai.