Poolside (@poolsideai) / X

Poolside

144 posts

Poolside

@poolsideai

We build models for agentic coding and long-horizon tasks. Try Laguna: poolside.ai

Joined May 2023

Pinned
Poolside
@poolsideai
Jul 2
Today we’re releasing Laguna XS 2.1. It’s a small upgrade to the Laguna XS.2 model, the same 33B total / 3B active MoE and stronger results on multilingual coding and terminal-style tasks. Available now on @huggingface, @OpenRouter, and via Poolside API.
59K
Poolside reposted
Cline
@cline
Jul 3
New free model in Cline! Laguna M.1 by Poolside. Speedy 225B total parameter model with 256k context, built for agentic coding and long-horizon work. Use the model id: poolside/laguna-m.1:free
17K
Poolside reposted
Baseten
@baseten
Jul 2
We're thrilled to power Laguna XS 2.1 from Poolside via the Baseten Frontier Gateway! Laguna XS 2.1 supports a 256K context window and is a strong default when you want a small, efficient model for agentic coding and long-horizon software tasks.
Poolside
@poolsideai
Jul 2
Today we’re releasing Laguna XS 2.1. It’s a small upgrade to the Laguna XS.2 model, the same 33B total / 3B active MoE and stronger results on multilingual coding and terminal-style tasks. Available now on @huggingface, @OpenRouter, and via Poolside API.
3.5K
Poolside reposted
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTex
Jul 2
Halfway to its Qwen 3.6 peer (I choose to ignore Terminal Bench). Presumably they, too, have learned the Dao of continuous RL gains. If this is their GLM 5.0 to 5.1 moment, I am optimistic about the next one
Poolside
@poolsideai
Jul 2
Today we’re releasing Laguna XS 2.1. It’s a small upgrade to the Laguna XS.2 model, the same 33B total / 3B active MoE and stronger results on multilingual coding and terminal-style tasks. Available now on @huggingface, @OpenRouter, and via Poolside API.
6.1K
Poolside reposted
LMSYS Org
@lmsysorg
Jul 2
🎉 Day-0 support for Laguna XS 2.1 from @poolsideai is now live in SGLang! This is a 33B total params MoE built for agentic coding and long-horizon work on your local machine. 1️⃣ Native interleaved thinking between tool calls, toggle per-request 2️⃣ Mixed SWA + global attention
Poolside
@poolsideai
Jul 2
Today we’re releasing Laguna XS 2.1. It’s a small upgrade to the Laguna XS.2 model, the same 33B total / 3B active MoE and stronger results on multilingual coding and terminal-style tasks. Available now on @huggingface, @OpenRouter, and via Poolside API.
3.4K
Poolside
@poolsideai
Jul 2
We are licensing Laguna XS 2.1 under OpenMDW-1.1. We are making this change to support open model distribution for the community. OpenMDW-1.1 is fully permissive and designed for models and related artifacts, giving developers and organizations a more consistent framework for
NVIDIA AI
@NVIDIAAI
May 28
We're adopting the Linux Foundation’s OpenMDW framework across our open model families. This helps make open model licensing simpler and more consistent at scale. A single legal framework across models, code, documentation, and data helps reduce friction for developers and
4.5K
Poolside
@poolsideai
Jul 2
Replying to @poolsideai @ollama and @atomic_chat_hq
For the best agent experience, try XS 2.1 in pool, our terminal-based coding agent. pool now supports OpenRouter login, so you can use it with Laguna XS 2.1, our other models, or the wider OpenRouter catalog. Just run pool login
GitHub - poolsideai/pool: pool is Poolside’s coding agent that runs in your terminal or integrates...
From github.com
1.3K
Poolside
@poolsideai
Jul 2
We’re working hard to release more capable models, and feedback from the community helps us make the next ones better. Join our Discord: discord.gg/sh2QJm2GA Learn more: poolside.ai/blog/introduci…
discord.com
Join the Poolside AI Discord Server!
The official Discord Server for Poolside.ai, welcome to the pool. | 172 members
1.1K
Poolside
@poolsideai
Jun 30
As you scale pre-training, things can get messy. Catch @robert_mchardy and @marah_i_abdin today at @aiDotEngineer World’s Fair for the behind-the-scenes of training Laguna models at Poolside! 11:10am–11:30am | Data Quality, Room 2024 (2nd floor)
1.4K
Poolside reposted
Arkadii
@ArkadiiBessonov
Jun 27
Three main ways to do FP8 in LLM pretraining — and they differ in mainly one thing: how the scale is attached. per-tensor vs blockwise vs MXFP8. Why pretraining has so much structure here: forward + backward is 3 matmuls (Fprop, Dgrad, Wgrad) across 3 tensor roles (weights,
34K
Poolside reposted
Interconnects
@interconnectsai
Jun 28
Artifacts 22: Zyphra, Cohere, Poolside, and others are expanding the breadth and diversity of the ecosystem. In this issue, a total of 30 models from may/june you should be aware of, from: NVIDIA @NVIDIAAI (3) Cohere @Cohere_Labs (2) Zhipu @Zai_org Zyphra @ZyphraAI (3) Poolside
23K
Poolside reposted
Marah Abdin
@marah_i_abdin
Jun 26
Looking forward to presenting with @robert_mchardy at AIEngineer World Fair in SF on Tuesday! on "The Messy Reality of Scale: Synthetic Data and Pre-Training at Poolside" ...we might also be finishing with a sneak peak on coming-soon Laguna model releases😉 @aiDotEngineer
Robert McHardy
@robert_mchardy
Jun 22
Looking forward to speaking at @aiDotEngineer next week together with @marah_i_abdin about synthetic data and pre-training at @poolsideai 🔥 I might show some base model evals of an upcoming models as well
25K
Poolside reposted
Robert McHardy
@robert_mchardy
Jun 22
Looking forward to speaking at @aiDotEngineer next week together with @marah_i_abdin about synthetic data and pre-training at @poolsideai 🔥 I might show some base model evals of an upcoming models as well
29K
Poolside reposted
Prince Canuma
@Prince_Canuma
Jun 18
Congrats to @poolsideai on the launch of Laguna M.1! We provide day-0 support on MLX 🚀 Get started: > uv pip install -U mlx-vlm
Poolside
@poolsideai
Jun 18
Today we’re releasing the weights for Laguna M.1, our most capable model to date, with a 256K context length. Both base and post-trained checkpoints are now available on Hugging Face under Apache 2.0.
14K