Laguna XS 2.1 33B-A3B — VRAM, szybkość i lokalna konfiguracja
Autor: Jakub Rusinowski · Ostatnia aktualizacja:
33B total parameters with only 3B activated per token, which is what lets a model scoring 70.9% on SWE-Bench Verified sit on a single 24GB card. Mixed sliding-window + global attention (3:1 across 40 layers), sigmoid per-head gating, FP8 KV cache, 256K context. Released under OpenMDW-1.1 — the Linux Foundation / NVIDIA permissive model-weights framework — not Apache 2.0, so check the terms if you are redistributing.
Laguna XS 2.1 33B-A3B potrzebuje około 21 GB VRAM przy Q4_K_M — skwantyzowane wagi plus narzut frameworka, bez pamięci podręcznej KV. Na Apple Silicon ta wartość pochodzi z pamięci zunifikowanej.
VRAM i szybkość według kwantyzacji
Liczone dla NVIDIA RTX 4090 (24 GB). Tylko wagi i narzut: architektura tego modelu nie jest opublikowana, więc nie doliczamy pamięci KV.
| Kwant. | Pamięć | VRAM | Szybkość (szac.) | Dopasowanie |
|---|---|---|---|---|
| Q2_K 2.63 bpw | 11.6 GB | ~210 tok/s | Mieści się | |
| Q3_K_M 3.41 bpw | 14.9 GB | ~194 tok/s | Mieści się | |
| Q4_K_M 4.83 bpw | 20.7 GB | ~171 tok/s | Mieści się | |
| Q5_K_M 5.67 bpw | 24.2 GB | ~24 tok/s | Częściowo w RAM | |
| Q6_K 6.56 bpw | 27.9 GB | ~22 tok/s | Częściowo w RAM | |
| Q8_0 8.50 bpw | 35.9 GB | ~20 tok/s | Częściowo w RAM | |
| F16 16.00 bpw | 66.8 GB | — | Za duży |
Czarny znacznik = dostępna pamięć na NVIDIA RTX 4090 (24 GB). Szacunki z modelu przepustowości pamięci opisanego na stronie metodologii. Kalkulator VRAM dla Laguna XS 2.1 33B-A3B →
Uruchom Laguna XS 2.1 33B-A3B
Najtańsze GPU z naszego katalogu, które uruchomi Laguna XS 2.1 33B-A3B, to AMD Radeon RX 7900 XTX (24 GB).
lub porównaj w Vast.ai od $0.35/godz. (typical low · varies)
Jako partner Amazon zarabiamy na kwalifikujących się zakupach. Linki do chmurowych GPU są linkami polecającymi — możemy otrzymać prowizję bez dodatkowych kosztów dla Ciebie.
Jak uruchomić Laguna XS 2.1 33B-A3B
Zainstaluj Ollama, a potem uruchom:
ollama run poolside-lagunaSpecyfikacja
Preview — The model is released, but these specs are thin or rest on a single source. Individual fields may be wrong.
Preview — The model is released, but these specs are thin or rest on a single source. Individual fields may be wrong.
- Parametry
- 33 Billion (3B active)
- Okno kontekstu
- 262,144
- Architektura
- Mixture-of-Experts (sliding-window + global attention, FP8 KV cache)
- Dostawca
- Poolside
- Licencja
- OpenMDW-1.1
- Practical deployment
- Single consumer GPU
- Określono przy
- Q4_K_M
- RAM systemowy
- 32 GB
- Rekord zaktualizowany
- 2026-08-15
Commercial use permitted. No usage restrictions beyond attribution.
Jakość i zastosowania
Wyniki opublikowane przez autorów modelu lub niezależnego oceniającego — jakość, nie przepustowość, i nie mierzone przez nas.
| Benchmark | Wynik | Źródło |
|---|---|---|
| SWE-Bench Verified | 70.9 / 100 % | opublikowane · https://poolside.ai/blog/introducing-laguna-xs-2-1 |
| SWE-bench Multilingual | 63.1 / 100 % | opublikowane · https://poolside.ai/blog/introducing-laguna-xs-2-1 |
Czy uruchomię Laguna XS 2.1 33B-A3B na moim GPU?
- Poolside Laguna XS 2.1 na AMD Radeon RX 7800 XT
- Poolside Laguna XS 2.1 na AMD Radeon RX 7900 XT
- Poolside Laguna XS 2.1 na AMD Radeon RX 7900 XTX
- Poolside Laguna XS 2.1 na AMD Radeon RX 9060 XT 16GB
- Poolside Laguna XS 2.1 na AMD Radeon RX 9060 XT 8GB
- Poolside Laguna XS 2.1 na AMD Radeon RX 9070
- Poolside Laguna XS 2.1 na AMD Radeon RX 9070 XT
- Poolside Laguna XS 2.1 na Apple M1
- Poolside Laguna XS 2.1 na Apple M1 Pro
- Poolside Laguna XS 2.1 na Apple M2
- Poolside Laguna XS 2.1 na Apple M2 Pro
- Poolside Laguna XS 2.1 na Apple M3
- Poolside Laguna XS 2.1 na Apple M3 Pro
- Poolside Laguna XS 2.1 na Apple M4
- Poolside Laguna XS 2.1 na Apple M4 Pro
- Poolside Laguna XS 2.1 na Apple M5
- Poolside Laguna XS 2.1 na Intel Arc B570
- Poolside Laguna XS 2.1 na Intel Arc B580
- Poolside Laguna XS 2.1 na NVIDIA GeForce RTX 3060 (12GB)
- Poolside Laguna XS 2.1 na NVIDIA GeForce RTX 3070
- Poolside Laguna XS 2.1 na NVIDIA GeForce RTX 3070 Ti
- Poolside Laguna XS 2.1 na NVIDIA GeForce RTX 3080 (10GB)
- Poolside Laguna XS 2.1 na NVIDIA GeForce RTX 3090
- Poolside Laguna XS 2.1 na NVIDIA GeForce RTX 4060
- Poolside Laguna XS 2.1 na NVIDIA GeForce RTX 4060 Ti 16GB
- Poolside Laguna XS 2.1 na NVIDIA GeForce RTX 4070
- Poolside Laguna XS 2.1 na NVIDIA GeForce RTX 4070 Super
- Poolside Laguna XS 2.1 na NVIDIA GeForce RTX 4070 Ti
- Poolside Laguna XS 2.1 na NVIDIA GeForce RTX 4070 Ti Super
- Poolside Laguna XS 2.1 na NVIDIA GeForce RTX 4080
- Poolside Laguna XS 2.1 na NVIDIA GeForce RTX 4080 Super
- Poolside Laguna XS 2.1 na NVIDIA GeForce RTX 4090
- Poolside Laguna XS 2.1 na NVIDIA GeForce RTX 5060
- Poolside Laguna XS 2.1 na NVIDIA GeForce RTX 5060 Ti 16GB
- Poolside Laguna XS 2.1 na NVIDIA GeForce RTX 5060 Ti 8GB
- Poolside Laguna XS 2.1 na NVIDIA GeForce RTX 5070
- Poolside Laguna XS 2.1 na NVIDIA GeForce RTX 5070 Ti
- Poolside Laguna XS 2.1 na NVIDIA GeForce RTX 5080
- Poolside Laguna XS 2.1 na NVIDIA GeForce RTX 5090
- Poolside Laguna XS 2.1 na NVIDIA L40S
- Poolside Laguna XS 2.1 na NVIDIA RTX 6000 Ada Generation
Laguna XS 2.1 33B-A3B — najczęściej zadawane pytania
How much VRAM does Laguna XS 2.1 33B-A3B need?
About 21 GB at Q4_K_M — quantized weights plus framework overhead, before any KV cache. The cache grows with context length and is added on top; the table above folds it in. Apple Silicon counts unified memory toward the same figure.
Does Laguna XS 2.1 33B-A3B run on an RTX 4090 (24 GB)?
Yes. Laguna XS 2.1 33B-A3B needs about 21 GB at Q4_K_M, inside a 24 GB card, at an estimated 171 tokens/sec.
How do I run Laguna XS 2.1 33B-A3B locally?
Install Ollama and run `ollama run poolside-laguna`. That pulls the weights and starts a local OpenAI-compatible endpoint; after the download nothing leaves the machine.