118B-A8B Poolside Laguna
Poolside's agentic coding model under its OpenMDW-1.1 licence: 117.6B parameters with about 8B active per token and a 1M-token window. Only 12 of its 48 layers keep a cache that grows with context; the rest use a 512-token sliding window. At Q4_K_M (ggml-org's 67.7 GB file) it needs about 71.0 GB at 4K, which is comfortable on a 128 GB DGX Spark or Mac, and the full 1M window adds only about 53 GB more. Poolside's own Q4_K_M file is far larger (96.0 GB), so check which file you download. llama.cpp runs it as laguna. Sizes come from real files; no per-level quality loss has been published.
1049K
Max Context
2
Quant Variants
GGUF Q8_0
Best Quality
—
Perplexity rise (best build)
Quantization Variants #
Per-quant VRAM, quality loss, and inference speed on RTX 4090
Measured = run on this site · Estimated = not run here: calculated, or a figure this site has not verified · Community = public reports
| Format | Level | BPW | VRAM | PPL Loss | Speed | Source | Actions |
|---|---|---|---|---|---|---|---|
| GGUF | Q4_K_M | 4.6 | 71.0 GB | — | — | Estimated | |
| GGUF | Q8_0 | 8.5 | 130.9 GB | — | — | Estimated |
Which file each size is for
A file size is not a memory figure: the VRAM column is the calculator's total at 4K context — weights plus KV cache plus a 10% activation buffer. Units differ on purpose: file sizes are decimal GB, as Hugging Face lists them; the VRAM column is GiB (1024³ bytes) with 2% added to the weights, so 7.12 GB of file counts as about 6.8 GiB there. A vision projector (mmproj), where the repo ships one, is a separate download and is in neither number.
- GGUF Q4_K_M[ggml-org/Laguna-S-2.1-GGUF](https://huggingface.co/ggml-org/Laguna-S-2.1-GGUF) · Laguna-S-2.1-Q4_K_M.gguf · 67.66 GB (67,661,639,264 bytes) · listing read 2026-10-23
- GGUF Q8_0[ggml-org/Laguna-S-2.1-GGUF](https://huggingface.co/ggml-org/Laguna-S-2.1-GGUF) · Laguna-S-2.1-Q8_0.gguf · 125.02 GB (125,022,858,848 bytes) · listing read 2026-10-23
Similar models #
[Compare with GPT-OSS 120B](https://quantized.uk/tools/compare/?a=laguna-s-2-1-118b-a8b&b=gpt-oss-120b)
[117B MoEPro GPUMac / Apple Silicon62.7 GBmin VRAM+0.0%perplexity rise](https://quantized.uk/quant-hub/gpt-oss-120b/)
GPT-OSS 120B
OpenAI GPT-OSS
The big GPT-OSS (117B total / 5.1B active). Native MXFP4 checkpoint is ~61GB — fits one 80GB card or a 128GB unified-memory Mac. Partial offload on 24GB consumer cards is slow but works.
[106B MoEPro GPUMac / Apple Silicon37.2 GBmin VRAM+2.4%perplexity rise](https://quantized.uk/quant-hub/glm-4.5-air/)
### GLM-4.5-Air
Zhipu GLM-4.5
Zhipu's agentic/reasoning MoE (106B total / 12B active). Q4 ~64GB — the sweet spot is a 96GB+ unified Mac or 2× 48GB cards. Strong tool-calling for its class.
[125B-A6BMac / Apple SiliconPro GPU79.3 GBmin VRAM—perplexity rise](https://quantized.uk/quant-hub/qwen3-8-flash-next/)
### Qwen3.8-Flash-Next
Qwen3.8
Qwen's preview of its next architecture: a 125B mixture-of-experts model with about 6B parameters active per token. Released in August, it also carries a 51B n-gram lookup table. llama.cpp reads that table lazily from RAM or disk, so it never needs VRAM. It does need disk space and is bundled inside the GGUF files, which is why they are far larger than the sizes below. Only 12 of its 48 layers keep a growing KV cache, so even the full 256K window adds about 7.5 GB. At Q4_K_M it needs about 79 GB at 4K context, comfortable on a 128 GB Mac or DGX Spark; at 256K it needs about 87 GB, still comfortable on the Spark but tight on the Mac, whose GPU gets about three quarters of its memory. A single 24 or 32 GB card runs it only with experts offloaded to system RAM. llama.cpp supports it as qwen4exp. Sizes use the calculator's generic rates, since no file size isolates the model from the table, and no per-level quality loss has been published.
816B-A42BMac / Apple SiliconPro GPU318.9 GBmin VRAM—perplexity rise
Muse Spark Open 816B-A42B
Meta Muse
Meta's largest open-weight model, released in October under the Muse Model License. It is an 816B mixture-of-experts model with about 42B parameters active per token, built for coding and agent work, with image and audio input. It is a single-machine model only at its smallest build. At Q2_K (ggml-org's 304 GB file) it needs about 319 GB at 4K context and still only 327 GB at 128K, because just 16 of its 62 layers keep a growing cache — comfortable on a 512 GB Mac Studio and tight across four 80 GB H100s. The full 1M window brings it to about 389 GB. Q4_K_S (470 GB) and Q8_0 (872 GB) fit no single machine in this index. The vision and audio encoders ship as separate files of 2–4 GB and are not included. Sizes come from the published GGUF files; Meta's benchmark claims are its own and no per-level quality loss has been published. The licence is not a standard open-source one: Meta's repository is gated, so downloads need an accepted access request; products built on it must show "Built with Muse Spark Open"; and a company with over $20M in annual revenue must sign a separate agreement with Meta before offering it as an API service.
Running Laguna S 2.1 118B-A8B locally #
At Q4_K_M and 4K of context, Laguna S 2.1 118B-A8B needs about 71.0 GB — 64.2 GB of weights, 0.33 GB of KV cache and a 6.5 GB activation buffer. The smallest cards in this index that clear that comfortably are 128 GB ones — 4 of them, from the DGX Spark 128G up — and 8 of the 78 cards here do. These are calculated figures, not measurements: the estimate stops counting a card as comfortable at 88% of its VRAM, which is roughly the room a desktop session needs.
What longer context costs #
Going from 4K to 32K adds about 1.31 GB, taking the total to 72.5 GB. That is gentler than the parameter count suggests: only 12 of 48 layers keep a KV cache that grows with context, and 36 more cap theirs at a sliding window, so the cache scales at a fraction of the usual rate. The model's native window is 1M; holding all of it at Q4_K_M would need about 124 GB.
Which build to download #
This index only tracks Laguna S 2.1 118B-A8B in GGUF, across 2 levels (Q4_K_M, Q8_0). No per-level perplexity sweep has been published for this model, so the quality column is empty rather than estimated — within one model, more bits per weight is the only ordering the data supports. No throughput has been measured for this model on this site's hardware. GGUF runs on llama.cpp and Ollama across NVIDIA, AMD and Apple silicon; AWQ and GPTQ target vLLM on CUDA and ROCm; EXL2 is ExLlamaV2 — CUDA only, and now archived, so local use rather than serving.
Common questions #
- How much VRAM does Laguna S 2.1 118B-A8B need?
- About 71.0 GB at Q4_K_M with 4K of context and batch 1, rising to roughly 72.5 GB at 32K. That figure is weights plus KV cache plus a 10% activation buffer, calculated from the model's architecture rather than measured on a card.
- Will Laguna S 2.1 118B-A8B run on a 128GB GPU?
- Yes — at Q4_K_M and 4K context it needs about 71.0 GB, which leaves 57.0 GB spare on a DGX Spark 128G, or on any of the other 128 GB cards here (4 in all). That is the smallest size in this index that clears it comfortably; 8 of 78 do.
- Which quantization of Laguna S 2.1 118B-A8B should I use?
- No published quality comparison exists for this model, so pick by footprint: Q4_K_M is the level most people run, and a higher bits-per-weight level is more faithful. All 2 levels in the index are Q4_K_M, Q8_0. Note this is a mixture-of-experts model — all parameters must be resident even though only a fraction are active per token, so the memory cost follows the total, not the active count.
Where this model fits #
Sized at Q4_K_M with a 4K context window, smallest card first. Comfortable means the estimate uses at most 88% of the memory.
- Runs comfortably on
- DGX Spark 128G128GB · +57.0Mac M3 Max 128G128GB · +25.0Mac M4 Max 128G128GB · +25.0Mac M5 Max 128G128GB · +25.0Mac M2 Ultra 192G192GB · +73.0
- Tight but possible
- A100 80G80GBH100 80G80GBMac M2 Max 96G96GB
- Just misses
- DGX Spark 64G+7.0GB overA40 48G+23.0GB over
- Ships in
- GGUF
- Try it yourself
- Size it in the calculatorGenerate the run command
How to actually run this #
Deployment guides for this model and this class of hardware.
This page's figures change when the model or the runtime does.Last updated 2026-10-23 RSS → /feed.xml