from openai import OpenAI
import os
client = OpenAI(
api_key=os.getenv("POOLSIDE_API_KEY"),
base_url="https://inference.poolside.ai/v1"
)
response = client.chat.completions.create(
model="poolside/laguna-s-2.1",
messages=[{
"role": "user",
"content": "What are channels in Go?"
}]
)Pulse. Real-time data from our models usage.
Models usage metrics, refreshed daily from OpenRouter and Hugging Face.
26.5 trillion tokens processed across all surfaces since our first launch.
Via OpenRouter
Other channels
- 89% Via OpenRouter, last 30 days
- 7.2% Direct API, last 30 days
- 3.3% Other gateways, last 30 days
- OpenCode Top surface, last 30 days
Stats for all models.
24.3 trillion tokens processed on OpenRouter since our first launch.
- 4.9 trillion Tokens last 30 days
- 1.1 trillion Tokens last 7 days
- 161 billion Tokens last day
- 329 billion Record daily tokens (4 Aug 2026)
Stats for all models.
Top models on OpenRouter last week
Tokens over the last 7 days. Arrows show places Laguna models moved vs. the previous week. Source: OpenRouter (openrouter.ai/rankings), as of 18 Oct 2026 (updated 12:10 UTC). View open models leaderboard
2.4M model weight downloads on Hugging Face since our first launch.
Download weights- 1.9M Downloads last 30 days
- 291.4k Downloads last 7 days
- 4.8k Downloads last day
- 21.2k Avg. downloads per day
Stats for all models.
Get started with Laguna. Free to use for a limited time.
$ curl -fsSL https://downloads.poolside.ai/pool/install.sh | sh
poolv1.0.6
• Connected to agent server · Laguna S 2.1
› Audit the checkout flow for race conditions
● Thinking
⎿ Tracing order creation and transaction boundaries
● Read(src/checkout/createOrder.ts)
● Edit(src/checkout/createOrder.ts)
● Bash(pnpm test checkout)
⎿ ✓ 12 tests passed in 284ms
⠹ Working… (12s)
Laguna in your terminal with pool.
$ docker run --gpus all poolside/laguna
NVIDIA DGX Spark · GB10 Grace Blackwell
128 GB unified memory · CUDA device ready
Loading poolside/laguna-s-2.1
Precision FP4 · Context 131,072
Starting OpenAI-compatible inference server...
✓ Ready at http://localhost:8000/v1
Run Laguna locally.
Work with Poolside Desktop Assistant.
Our partners
Discover how we built our models.
-
The Model Factory series
A six-part blog series on how we train, evaluate, and iterate on foundation models. From GPU-to-GPU weight transfers and automated architecture ablations to reinforcement learning from code execution at scale. Laguna S 2.1: a small model that refuses to give upNew
Today we're releasing Laguna S 2.1, a 118B-parameter model (8B active) with a 1M-token context window, in thinking and no-thinking modes. On long-horizon coding benchmarks it holds its own against models many times its size.Laguna M.1/XS.2 Technical Report
We present LAGUNA M.1 and LAGUNA XS.2, two Mixture-of-Experts foundation models built for long-horizon, agentic coding: M.1 has 225.8B total parameters (23.4B activated per token) and XS.2 has 33.4B total (3B activated).Introducing Laguna XS.2 and Laguna M.1
We’re releasing two foundation models and two products into preview today.