Laguna S 2.1

(poolside.ai)

159 points | by rexledesma 4 hours ago

16 comments

  • Lwerewolf 1 hour ago

    Testing it now. At the very least, competitive with DS4-Flash indeed. On my small (and per Sol's words, _very_ semantically dense) C test codebase, it found things that only gpt-5.2 managed to find back in the day, but also made a stupidly incorrect initial observation that a memfd_create()/mmap was used for IPC (funnily enough - sol missed that as well in its review, until I pointed it out). Re: the claims vs deepseek v4 - both flash and pro are expected to get a "general availability" release very soon (i.e. well-"post-trained"), so things can change in a... well, flash, as per usual in the current environment.

    Anyways, keep 'em coming.

    • ilc 1 hour ago

      What harness/quant did you use for testing?

      • Lwerewolf 1 hour ago

        nvfp4 mlx, literally barebones pi.

        edit: on bigger tests, got it to loop pretty easily unfortunately, probably local settings.

        • sosodev 33 minutes ago

          What inference server are you using? They have a custom branch for llama.cpp, but I wouldn't be surprised at all if it still needs fixing.

    • mft_ 1 hour ago

      Looks impressive, and this size fits achievable home hardware.

      That said, if someone would kindly quantise this down for the 64GB paupers, that would be appreciated. (I know there’s likely degradation, but some people reported good results with a 2 bit version of Qwen 3.5 122B, and this is starting from a higher point. Would be interesting to try, at least.)

      Edit: someone in the process of doing so: https://huggingface.co/vcruz305/Laguna-S-2.1-GGUF

    • river_otter 1 hour ago

      Hey, this model is not a joke! Exciting, we already got a usable PR of work out of it.

      https://github.com/mozilla-ai/otari/pull/348

      • kamranjon 2 hours ago

        Whoa whoa whoa, 118b params, 8b active MOE, long context reasoning, open weights - music to my ears. Hadn't heard of this lab before but I am very excited, will definitely try this out tomorrow - this is a real sweet spot I think in terms of model size and performance.

        • svclaws 1 hour ago

          If the numbers are legitimate then our prayers have been heard

        • mchusma 1 hour ago

          Incredible. This is definitely the launch of the day. Just crushing Google's releases.

          The pricing here is incredible. This is the first US release that's competitive with DeepSeek V4 Flash. Very excited about this.

          • benjiro29 29 minutes ago

            !! Be careful when testing the model.

            A lot of people are testing it, and reporting disappointed results / benchmaxxxing claim. But do not realize that thinking has a issue with the default configuration.

            Important - make sure that THINKING is enabled. By default it wasn't although I was passing the flag --default-chat-template-kwargs '{"enable_thinking": true}' in vllm recipe. The generation_config.json file that is included has by default max_new_tokens as 32k which seems to be cutting off thinking altogether so increase it. At first I was very disappointed with the output I was seeing, but once thinking is enabled, the code quality seems to be MUCH better. More real world testing to be done.

            https://www.reddit.com/r/LocalLLaMA/comments/1v2pg99/laguna_...

            • Iolaum 2 hours ago

              Model Looks amazing!

              Even more important, subjectively, is that this model will run very well on Strix Halo (e.g. Framework Desktop), DGX Spark kinds of devices. Looking forward to Unsloth dynamic mtp quants.

              P.S. Looking at the HF release they already offer Q4_K_M and DFlash drafter for speculative decoding!

              • verdverm 1 hour ago

                I hope all models going forward come with a dflash drafter so we don't have to train one up separately.

              • loolhahalmao 15 minutes ago

                happy the US has some counterweights to the Chinese labs, just need about half a dozen more.

                • SwellJoe 2 hours ago

                  This is exactly the kind of model that's been needed in the middle. Realistically self-hosted, Good Enough intelligence, MoE so it's fast on limited bandwidth systems like Strix Halo and DGX Spark.

                  For a while there's been nothing to run on my Strix Halo that's notably better than what I can run on my dual 32GB GPU desktop (Gemma 4 or Qwen 3.6 dense models), but this seems likely to be the step up in size that actually works better than those.

                  • docheinestages 42 minutes ago

                    Any estimates of the performance (prompt processing and decoding tokens/s) on consumer hardware like Macbook Pro M-series?

                    • river_otter 2 hours ago

                      I love this. Is it possible to give a feel of how this stacks up to the good old Opus 4.5 in coding quality? For me that was the turning point where agentic coding in Claude Code etc became usable. Have we hit that threshold?

                      • megavon 1 hour ago

                        Having played with it for like 3 hours now....I'm probably moving from CC to this

                        • river_otter 1 hour ago

                          I am about 1 hour into using it with pi.dev. Do you have thinking on high? It is doing good but at one point i had to stop it and say 'you're overthinking this' haha

                          • megavon 1 hour ago

                            Yes full send mode on thinking. I have moved on from watching my agents and I don't really care how it thinks. I look at the end result and so far this thing has been blowing me away. No way this is as good as it is this small and fast. Outside Fable, this might be the best thing I've ever used.

                            • kamranjon 1 hour ago

                              What quant are you using?

                          • fingerprinter 47 minutes ago

                            One hour in, no more Codex for me. This thing rips.

                        • megavon 3 hours ago

                          This is INSANE. How did they do this?

                          • eisokant 2 hours ago

                            "What we've done in this model is not necessarily add more intelligence, but improve the behaviors that lead to a more capable model: more verification, less taking things for granted, not declaring victory early, and being more persistent.”

                            +

                            https://poolside.ai/assets/laguna/laguna-m1-xs2-technical-re...

                            • Lwerewolf 1 hour ago

                              Almost like a built-in heavyweight harness.

                            • kamranjon 1 hour ago

                              "It went from the start of training to launch in under nine weeks..."

                              This is pretty impressive.

                            • tosh 3 hours ago

                              open weights and

                              similar performance to deepseek v4, inkling at size of nemotron 3 super (!)

                              • iraldir 3 hours ago

                                Amazing model at this size if true, that's quite crazy!

                                • carimura 54 minutes ago

                                  Congrats Poolside team!!

                                  • danr4 28 minutes ago

                                    holy shit its accelerating fast