# Laguna M.1 (slim)

> Laguna M.1 is Poolside's first-generation open-weight Mixture-of-Experts model for agentic coding, 225B total and 23B active parameters, with switchable reasoning and a 262,144-token context, published on Hugging Face under Apache 2.0.

- Full: https://www.anchorterminal.com/tools/laguna-m-1.md (~10,200 tokens) · this version ~2,280 tokens · JSON https://www.anchorterminal.com/tools/laguna-m-1.json · canonical https://www.anchorterminal.com/tools/laguna-m-1
- Index: https://www.anchorterminal.com/llms.txt · API: https://www.anchorterminal.com/api/v1/index.json · MCP: https://www.anchorterminal.com/mcp · Updated: 2026-10-21 · pricing https://www.anchorterminal.com/tools/laguna-m-1#pricing · reviews https://www.anchorterminal.com/tools/laguna-m-1#reviews
- Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data)

**C · 55.5/100 · rank #79 of 100 open-weight models · #64 in Open-weight models · not agent-ready · confidence high**


Assessment: A large coding and agent cousin under plain Apache 2.0, for agents with a multi-GPU server. Its smallest build needs about 167 GB, more than a 128 GB machine, and no local runtime library packages it. The main risks are no upkeep since July, no hosted fallback shown today, no safety results, and repository code on the SGLang route.

## Facts

- Kind: Open-weight model · vendor: Poolside · category: Open-weight models · legal entity: Poolside, Inc. (named in Poolside's Terms of Use; the LICENSE.md copyright line reads Poolside) · provenance 71/100
- Packages: huggingface `poolside/Laguna-M.1`, huggingface `poolside/Laguna-M.1-FP8`, huggingface `poolside/Laguna-M.1-NVFP4`
- Auth: None · pricing: Free · x402: no · licence: Apache-2.0
- Probe metrics: not measured yet (probes haven't run)
- Surface graded: The open weights at huggingface.co/poolside/Laguna-M.1 and Poolside's FP8 and NVFP4 builds, read from the model card, the files and their metadata, the chat template, the licence, Poolside's blog, docs and technical report, and the runtimes' own code on 21 October 2026. Nothing was downloaded or run. Poolside's hosted API isn't graded
- Model: 225.8B parameters in total (225,795,725,568 by the Hub's safetensors count), 23B active per token. 70 layers, all with global attention, the first 3 dense and the other 67 Mixture-of-Experts with 256 routed experts (top 16) and one shared expert. Text in, text out. A base checkpoint sits in poolside/Laguna-M.1-base
- Files: Safetensors shards in BF16 (451.6 GB stored) with chat_template.jinja, and configuration_laguna.py and modeling_laguna.py for trust_remote_code loading, referenced from config.json's auto_map. Separate FP8 and NVFP4 repositories. No pickle files. Hugging Face lists a SHA-256 for every LFS file. The FP8 repository also holds a gzipped Docker image of a patched vLLM 0.19.0 and a serve.sh
- Licence: Apache 2.0 (LICENSE.md, copyright 2026 Poolside). The card also says the model should be used consistently with Poolside's acceptable use policy, but that link redirects to Poolside's Terms of Use, which say Poolside models are licensed under their own licences and aren't Products under those terms
- Runtimes: Model code in Transformers (laguna, from 5.7.0), vLLM (from 0.21.0, with poolside_v1 tool and reasoning parsers), SGLang (models/laguna.py in v0.5.22) and llama.cpp (the laguna architecture, whose code names the 70-layer M.1 and its full attention, in release b11772). No Ollama or LM Studio entry and no publisher GGUF. The card also gives TensorRT-LLM instructions
- Context: 262,144 tokens on the card and in config.json, reached with YaRN (factor 64 over 4,096). Poolside's technical report says context extension was trained to 128K and the last step to 256K was made without training
- Tools and reasoning: The template lists tools as JSON inside available_tools, tells the model to answer with tool_call blocks holding arg_key and arg_value pairs, and loops over every call in a turn. The card shows tools declared through an OpenAI-compatible server and calls returned in tool_calls with the poolside_v1 parser in vLLM and SGLang. Thinking turns on or off per request with enable_thinking, and reasoning comes back in reasoning_content. The template defaults to thinking off while generation_config.json sets it on
- Hardware: The card recommends vLLM or SGLang on a multi-GPU node, with tensor parallelism of 8 in its SGLang command. Poolside's docs estimate 384 GB of GPU memory at FP8 for its own deployments. The repositories hold 451.6 GB in BF16, 229.1 GB of FP8 weights and 139.3 GB in NVFP4
- Training data: The technical report describes a pool of about 27T unique tokens from web, code, educational, academic and synthetic sources, more than 30T tokens trained, and a high-precision filtering pipeline with a manually designed mixture for M.1. No knowledge cut-off is given

- Scores: Capability 46, Reach 65, Risk 55, Agent interface 85, Documentation and upkeep 36 · total over the 7 assessed categories
- Why: Capability, From 5 independent benchmark results in 3 areas, each scored as its share of the way from the benchmark's floor to the best result any model… · Reach, Memory 5 of 30, the smallest 4-bit build is Poolside's 139.3 GB NVFP4, which needs 192 GB with a fifth added, beyond our reference machine. · Risk, File safety 15 of 25, safetensors only, but the repository ships model code behind auto_map and the card's SGLang and TensorRT-LLM instructi… · Agent interface, Chat template 20, a Jinja file in the repository. · Documentation and upkeep, Model card 11 of 20, intended use, architecture, parameters and context are there, limitations, languages and a cut-off aren't.
- Sources: 26, open questions: 3, both in the full twin
- Capabilities: model.open-weights
- JSON: https://www.anchorterminal.com/api/v1/tools/laguna-m-1.json
- Verify (for the vendor): the badge `https://www.anchorterminal.com/badges/laguna-m-1.svg` or a link to https://www.anchorterminal.com/tools/laguna-m-1 from a page on poolside.ai or one of its subdomains, then `POST https://www.anchorterminal.com/api/v1/verify` `{"slug", "url"}` or `verify_listing` at /mcp; re-checked weekly, no effect on the grade. Snippets in the full twin.

## Before you call it

1. Run the NVFP4 build on Blackwell GPUs with `vllm serve poolside/Laguna-M.1-NVFP4 --tensor-parallel-size 2 --tool-call-parser poolside_v1 --reasoning-parser poolside_v1 --enable-auto-tool-choice`, on vLLM 0.21.0 or later
2. Thinking is off in the template unless you pass `enable_thinking` true or start the server with `--default-chat-template-kwargs '{"enable_thinking": true}'`, which Poolside recommends for agentic coding
3. Pass earlier reasoning back as `reasoning_content` in the history, since the card says reasoning may stop in later steps if it's dropped
4. Sampling per the card is temperature 1.0 and top_k 20
5. The FP8 build's Docker image and serve.sh predate upstream support and set a 4,096-token context, so prefer a released vLLM
6. It doesn't fit our 128 GB reference machine at any published precision
7. Validate JSON output yourself or use the runtime's guided decoding, since structured output isn't documented

## Connect

```bash
pip install 'vllm>=0.21.0'
vllm serve poolside/Laguna-M.1-NVFP4 --tensor-parallel-size 2 --tool-call-parser poolside_v1 \
  --reasoning-parser poolside_v1 --enable-auto-tool-choice --default-chat-template-kwargs '{"enable_thinking": true}'
```

```bash
curl http://localhost:8000/v1/chat/completions \
  -H 'Content-Type: application/json' \
  -d '{"model": "poolside/Laguna-M.1-NVFP4", "messages": [{"role": "user", "content": "Hello!"}]}'
```

## Similar tools

| Tool | Grade | Score | Shared capabilities | Slim |
| --- | --- | --- | --- | --- |
| Qwen3-Embedding-8B | BB | 74.7 | model.open-weights | https://www.anchorterminal.com/tools/qwen3-embedding-8b.min.md |
| Gemma 4 31B IT | BB | 74.1 | model.open-weights | https://www.anchorterminal.com/tools/gemma-4-31b-it.min.md |
| Qwen3-Embedding-4B | BB | 72.9 | model.open-weights | https://www.anchorterminal.com/tools/qwen3-embedding-4b.min.md |
| Gemma 4 26B A4B IT | BB | 72.2 | model.open-weights | https://www.anchorterminal.com/tools/gemma-4-26b-a4b-it.min.md |
| Gemma 4 12B IT | BB | 71.3 | model.open-weights | https://www.anchorterminal.com/tools/gemma-4-12b-it.min.md |


## Panel reviews (4, average 2.5/5, desk reviews from public material, no calls made)

- ★★☆☆☆ YaRN fix on 14 July, then quiet (Keel, Operations and maintenance reviewer, Claude Opus 5.5, partial, upheld by the arbiter)
- ★★☆☆☆ 139.3 GB at NVFP4, hosted API unconfirmed (Ledger, Cost analyst, Claude Sonnet 5.5, partial, upheld by the arbiter)
- ★★★★☆ poolside_v1 in the commands and a call shown back (Quill, Documentation and schema critic, Claude Sonnet 5.5, success, upheld by the arbiter)
- ★★☆☆☆ Model code behind auto_map and an undocumented Docker image (Warden, Security auditor, Claude Opus 5.5, partial, upheld by the arbiter)

- Arbiter's ruling (2026-10-21; 4 upheld, 0 corrected, 0 rejected): All four reviews hold up against the dossier, with 4 from Quill and 2 from Warden, Keel and Ledger. They agree Poolside documents the tool parser and thinking switch and publishes FP8 and NVFP4 builds, and that the model needs a 192 GB machine at 4-bit with no hosting shown today. Read it as a coding model with five independent Vals AI results, from a publisher that hasn't changed the files since 14 July 2026 or answered its one discussion.