Model Follower.

Up to date

Rebuilt
2026-10-25 06:54 UTC
Release radar
117 on watch
Hub snapshot
2026-10-25
Re-checked
daily · every number sourced and dated

Models › Laguna family

Laguna-S-2.1

Hugging Face ↗ Follow changes

poolside out 3 months ago

118B parameters (10 of 256 experts per token), text in, text out, 1M tokens of context. 2 providers serve it.

Verified 2026-10-25 Rank #310 in LLMs · score 83.1 method v1 Sources 4 Downloads 30d 464.4k · #169 of 619 tracked

Can I run it?71GB at INT4Estimatefits 2×80 GBmethodCommunity: Laguna S 2.1: Our Benchmarks, VRAM and Which Version to Get →
What does it cost?$0/1M infree tier ⓘPoolside2026-10-24
Do people use it?464.4kdownloads, 30 days#169 of 619 we track · top 27%+7 likes, downloads +31% this week1,063 likesHugging Face2026-10-24

Worth reading

Picked by rule from 21 tracked pieces: the lab's own post, the most shared, a field report, the latest.

Laguna S 2.1 scored 40.4% in our agent harness, pool, not mini-swe-agent which DeepSWE's leaderboard uses.

From Laguna S 2.1 · poolside.ai · 2026-07-21
Most discussed · 416 HN pts

Laguna S 2.1

Poolside releases Laguna S 2.1, an agentic coding model with 118B total parameters and 8B activated parameters per token that launched in under nine weeks.

  • Laguna S 2.1 scored 40.4% in our agent harness, pool, not mini-swe-agent which DeepSWE's leaderboard uses.
poolside.ai 2026-07-21 416 HN pts · 88 comments 3.8k words sells inference

21 best of 21 tracked

The 4 picks above, plus these 17. Top is the radar's score: discussion, source authority and the lab's own posts, fading with age.

  1. ArticleLaguna XS.2 and M.1poolside.ai · 2026-04-28 · 101 HN pts
  2. ArticleLaguna XS 2.1 33B on a RTX 3090: 296 tok/s peak, 152 tok/s at 256K contextlucebox.com · 2026-07-11 · 20 HN pts
  3. DiscussionLaguna S 2.1:118B-a9B better than Qwen3.5:122B-a10B? So far, yesnews.ycombinator.com · 2026-08-06 · 4 HN pts
  4. ArticleLaguna XS.2 and Laguna M.1 by Poolsidepoolside.ai · 2026-04-28 · 5 HN pts
  5. Codeevidence/reports/laguna-s21-strix-halo-vulkan-benchmark.mdgithub.com · undated
  6. CodeBlackwellboy/laguna-s21-labgithub.com · undated
  7. ArticleInside Poolside Laguna S 2.1: The 118B-A8B Open-Weight MoE Built for 1M-Context Agentic Coding | Lobhari Technologieslobhari.in · 2026-08-20
  8. ArticleLaguna M.1/XS.2 Technical Report [pdf]poolside.ai · 2026-05-26
  9. ArticleLaguna S 2.1: Don't Let It Think - peasterpeaster.io · 2026-07-22
  10. ArticleWhat Is Laguna S 2.1? 118B-A8B Open Coding MoEkie.ai · 2026-07-22
  11. ArticleLaguna S 2.1 Coding Model Tested for Agent Workthevibefather.com · 2026-07-22
  12. ArticleWatch Tesla's Record-Breaking Laguna Seca Lapengadget.com · 2019-09-12
  13. ArticleLaguna XS.2: A Coding Model Built for Closed Environments, Now Open Weightsfirethering.com · 2026-04-29
  14. ArticleLaguna S 2.1: Our Benchmarks, VRAM and Which Version to Getmodeltested.com · undated
  15. ArticleLaguna S 2.1 · LLM Architectures Explainedllm-architectures-explained.vercel.app · undated
  16. VideoTesla Model S at Refuel 2013 (Laguna Seca)youtube.com · 2013-09-19
  17. DiscussionTesla is attempting lap record with a Model S Plaid+ at Laguna Seca Racewaytwitter.com · 2021-05-17

Where to run it

Live hosted endpoints, cheapest first, USD per million tokens. Ties break on output price, then provider name.

#Cheapest 2In /1MOut /1MContextQuantSeen
1 Poolside · free tier $0 $0 1M fp4 Oct 24
2 Featherless — — — — Oct 25

All 2 providers →

Where it's used

Hugging Face does not say who downloads a model or with what. These are the proxies: API providers listing it, quantized builds, fine-tunes/adapters/merges, and Spaces that load it.

Served by 2 API providers; heavily run locally (96 quantized builds).

  • As a service2Featherless, Poolside
  • On own hardware96quantized builds
  • Built on2322 fine-tunes · 1 adapters · 0 merges
  • In apps48Hugging Face Spaces

Top Spaces running it: poolside/Laguna-S-2.1 ♥ 29 · akhaliq/Laguna-S-2.1 ♥ 3 · Shantanupathak94/Ethrix-Ai ♥ 2 · johncenaqweewewwe/neuroResearch ♥ 1

Will it fit?

Weights only, an estimate: leave room for the KV cache. Method

Fits two 80 GB GPUs at INT4; too big for one

About 71 GB at INT4, 141 GB at FP8, 282 GB at BF16: weights only, so leave room for the KV cache. Every expert must be resident, so total parameters count.

16 GB card64 GB Mac2×80 GB8×80 GBINT4: about 71 GB of weights (estimate; KV cache and activations extra)INT471 GBFP8: about 141 GB of weights (estimate; KV cache and activations extra)FP8141 GBBF16: about 282 GB of weights (estimate; KV cache and activations extra)BF16282 GB
16 GB card64 GB Mac2×80 GB8×80 GBINT4: about 71 GB of weights (estimate; KV cache and activations extra)INT471 GBFP8: about 141 GB of weights (estimate; KV cache and activations extra)FP8141 GBBF16: about 282 GB of weights (estimate; KV cache and activations extra)BF16282 GB

Estimate params × bytes per param × 1.2 · how we estimate

The story so far

Weights, every dated article, and repo commits on one rail.

Most coverage came in the first two weeks

Teal is the weights (2026-07-13), amber triangles are 8 dated articles (each links to it), grey ticks are repo commits.

MayJunJulAugSepOct2026-08-19: 1 repo commit2026-07-27: 1 repo commit2026-07-24: 1 repo commit2026-07-22: 1 repo commit2026-07-21: 6 repo commitsLaguna S 2.1 · 2026-07-21Laguna XS.2 and M.1 · 2026-04-28Laguna XS 2.1 33B on a RTX 3090: 296 tok/s peak, 152 tok/s at 256K context · 2026-07-11Anybody tried Laguna S 2.1 (by Poolside)? · 2026-08-06Running Laguna S 2.1 locally on Apple Silicon: 52 tok/s with 38.5 GB peak memory · 2026-07-22Laguna S 2.1:118B-a9B better than Qwen3.5:122B-a10B? So far, yes · 2026-08-06Agentic AI at Two Different Scales: Nanbeige4.2-3B and Laguna S2.1 · 2026-07-31Laguna XS.2 and Laguna M.1 by Poolside · 2026-04-28Weights 2026-07-13 · 2026-07-13Weights 2026-07-13now · 2 providers
MayJunJulAugSepOct2026-08-19: 1 repo commit2026-07-27: 1 repo commit2026-07-24: 1 repo commit2026-07-22: 1 repo commit2026-07-21: 6 repo commitsLaguna S 2.1 · 2026-07-21Laguna XS.2 and M.1 · 2026-04-28Laguna XS 2.1 33B on a RTX 3090: 296 tok/s peak, 152 tok/s at 256K context · 2026-07-11Anybody tried Laguna S 2.1 (by Poolside)? · 2026-08-06Running Laguna S 2.1 locally on Apple Silicon: 52 tok/s with 38.5 GB peak memory · 2026-07-22Laguna S 2.1:118B-a9B better than Qwen3.5:122B-a10B? So far, yes · 2026-08-06Agentic AI at Two Different Scales: Nanbeige4.2-3B and Laguna S2.1 · 2026-07-31Laguna XS.2 and Laguna M.1 by Poolside · 2026-04-28Weights 2026-07-13 · 2026-07-13Weights 2026-07-13now · 2 providers

What people built on it

119 derivatives on the Hub. Which build to run, by runtime: the most downloaded in each class.

119 derivatives on the Hub, most of them quantized

From the Hugging Face model tree; quantized builds in teal, fine-tunes, merges and adapters in greys.

96 quantized (Hugging Face model tree)22 fine-tunes (Hugging Face model tree)1 adapter (Hugging Face model tree)96 quantized22 fine-tunes1 adapter
96 quantized (Hugging Face model tree)22 fine-tunes (Hugging Face model tree)1 adapter (Hugging Face model tree)96 quantized22 fine-tunes1 adapter
GGUF poolside/Laguna-S-2.1-GGUF for llama.cpp, Ollama, LM Studio 1.7M downloads MLX poolside/Laguna-S-2.1-NVFP4-mlx for Apple silicon 20k downloads NVFP4 poolside/Laguna-S-2.1-NVFP4 for vLLM, SGLang 1.2M downloads

All variants →

Reference

Everything else, folded: the full tables, every source.

Benchmarks

9 rows

Benchmarks 9 rows

measured third-party leaderboard rowauthor-reported number from the model card or eval files (9)No confidence intervals unless the source gave them; settings differ between rows, so compare within a benchmark, not across. Date = the benchmark's own date when it gives one, else when we read it.
BenchmarkValueSetting / notesReported byDate
author-reported DeepSWE · Write code with agents40.4% percent118B-A8Bmodel author2026-10-07
author-reported SWE Atlas (Codebase QnA)46.2% percent118B-A8Bmodel author2026-10-07
author-reported SWE-bench Multilingual · Write code with agents78.5% percent118B-A8Bmodel author2026-10-07
author-reported SWE-bench Pro · Write code with agents59.4% percent118B-A8Bmodel author2026-10-07
author-reported Terminal-Bench · Write code with agents70.2% percent118B-A8Bmodel author2026-10-07
author-reported Toolathlon · Call tools and APIs49.7% percent118B-A8Bmodel author2026-10-07
author-reported deep_swe (datacurve/deep-swe)40.4—Model Card2026-10-24
author-reported swe_bench_%_resolved (SWE-bench/SWE-bench_Multilingual)78.5—Model Card2026-10-24
author-reported SWE_Bench_Pro (ScaleAI/SWE-bench_Pro)59.4—Model Card2026-10-24

What the card says

3 cited sentences

Summary generated, cited

Laguna S 2.1 is a 118B parameter Mixture-of-Experts model designed for agentic coding and long-horizon work.[1][2] The architecture utilizes 256 routed experts and features a 1,048,576-token context window.[3][4] It is released under the OpenMDW-1.1 license.[5]

  1. parameters human
  2. “designed for agentic coding and long-horizon work”
  3. num experts
  4. “1,048,576-token context window”
  5. “OpenMDW-1.1 license”

Generated 2026-10-07 15:08 UTC by parasail-gemma-4-26b-a4b-it; each numeral points at the fact it rests on. Not a verdict.

Use-case evidence

5 use cases

What it is good at 5 use cases

Labels are evidence, not verdicts: measured a third-party leaderboard row for this model; author-reported a number the model author published in the card or eval files; structural a deterministic signal from the name, config or chat template, or a decision-model estimate. Each chip links to its source. All use cases.

Facts

22 fields

Listed in LLMs · MoE

Facts

Pipeline tagtext-generation
Librarytransformers
ArchitecturesLagunaForCausalLM
Model typelaguna
Dtypebfloat16
Params by dtypeBF16: 117,561,977,600
Size bandXXL
MoEyes
Experts256
Experts per token10
Context length1,048,576 (config.json max_position_embeddings)
Hub snapshot2026-10-25

Servability

vLLM · SGLang

Servability

vLLM: supported · LagunaForCausalLM · 7050ef881ff0835f6890a47db1859a24c65481d6SGLang: supported · LagunaForCausalLM · 773cb938fe9484a8f7e7e46fb1b1fe96412c59ed

vLLM registry (7050ef881ff0835f6890a47db1859a24c65481d6) · SGLang registry (773cb938fe9484a8f7e7e46fb1b1fe96412c59ed) · Matched by architecture class name against each project's registry on main; 'supported' means the architecture is registered, not that this checkpoint was tested.

Hardware fit

3 estimates

Hardware fit ESTIMATE

Weights-only memory for Laguna-S-2.1: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.

BF16 (2 bytes/param)282 GB · Multi-GPU node (<640 GB)
FP8 (1 byte/param)141 GB · Multi-GPU node (<640 GB)
INT4 (0.5 bytes/param)71 GB · Single high-end GPU (<80 GB)

Weights only (params × bytes/param × 1.2 overhead); KV cache and activations extra. MoE: all experts must be resident, so total parameters count.

All variants

40 variants

Lineage

Original model Family: poolside/Laguna 119 derivatives on the Hub

quantized: 96finetune: 22adapter: 1

Top variants (40 shown, ranked by downloads)

Relation
Quant
Size
40 rows
Variant Relation Quant Params Band License Author Likes Downloads
poolside/Laguna-S-2.1-GGUF quantized GGUF — — — poolside 178 1.7M
poolside/Laguna-S-2.1-NVFP4 quantized NVFP4 118B XXL openmdw-1.1 poolside 201 1.2M
Myric/Laguna-S-2.1-APEX-GGUF quantized GGUF — — openmdw-1.1 Myric 25 343k
unsloth/Laguna-S-2.1-GGUF quantized GGUF — — openmdw-1.1 unsloth 311 252.8k
poolside/Laguna-S-2.1-INT4 quantized INT4 118B XXL openmdw-1.1 poolside 28 250.7k
poolside/Laguna-S-2.1-FP8 quantized -FP8 118B XXL openmdw-1.1 poolside 28 128.3k
lmstudio-community/Laguna-S-2.1-GGUF quantized GGUF — — openmdw-1.1 lmstudio-community 2 95.1k
huihui-ai/Huihui-Laguna-S-2.1-abliterated-GGUF quantized GGUF — — openmdw-1.1 huihui-ai 14 41.4k
poolside/Laguna-S-2.1-DFlash finetune — 1.11B S other poolside 26 25.7k
poolside/Laguna-S-2.1-NVFP4-mlx quantized NVFP4 118B XXL openmdw-1.1 poolside 29 20k
bartowski/Laguna-S-2.1-GGUF quantized GGUF — — openmdw-1.1 bartowski 9 15.9k
cyankiwi/Laguna-S-2.1-AWQ-INT4 quantized AWQ 118B XXL openmdw-1.1 cyankiwi 4 16.1k
olka-fi/Laguna-S-2.1-MXFP4 quantized MXFP4 64.3B XL openmdw-1.1 olka-fi 3 10k
jcbtc/Laguna-S-2.1-Chadrock-ROCmFP4-StrixKVSpine-V4-GGUF quantized GGUF — — openmdw-1.1 jcbtc 37 8k
antirez/Laguna-S-2.1-GGUF quantized GGUF — — — antirez 11 8.9k
wimmmm/poolside-Laguna-S-2.1-GGUF quantized GGUF — — openmdw-1.1 wimmmm 1 8.7k
mlx-community/Laguna-S-2.1-oQ4e quantized — 118B XXL openmdw-1.1 mlx-community 10 7k
mlx-community/Laguna-S-2.1-oQ2e-fast quantized — 118B XXL openmdw-1.1 mlx-community 13 5.8k
AtomicChat/Laguna-S-2.1-GGUF quantized GGUF — — openmdw-1.1 AtomicChat 11 5.7k
mlx-community/Laguna-S-2.1-oQ2e quantized — 118B XXL openmdw-1.1 mlx-community 12 5.1k
mlx-community/Laguna-S-2.1-oQ6e quantized — 118B XXL openmdw-1.1 mlx-community 7 3.7k
kkuspa/Laguna-S-2.1-NVFP4-0804 quantized NVFP4 118B XXL openmdw-1.1 kkuspa 5 3.3k
Abiray/Laguna-S-2.1-IQ4_XS-Q4_K_M-GGUF quantized GGUF — — openmdw-1.1 Abiray 13 2.7k
TensorFold/Laguna-S-2.1-MLX-4bit quantized MLX 118B XXL openmdw-1.1 TensorFold 4 3.1k
dealignai/Laguna-S-2.1-CRACK-GGUF quantized GGUF — — openmdw-1.1 dealignai 17 2.4k
kingjones777/Laguna-S-2.1-ROCmFP4-STRIX_LEAN-GGUF quantized GGUF — — other kingjones777 4 2.9k
TobDeBer/Laguna-S-2.1-GGUF quantized GGUF — — openmdw-1.1 TobDeBer 1 2.8k
multimodalart/laguna-s-2.1-DFlash quantized GGUF — — openmdw-1.1 multimodalart 3 2.6k
mlx-community/Laguna-S-2.1-oQ3e-fast quantized — 118B XXL openmdw-1.1 mlx-community 10 2.2k
coolthor/Laguna-S-2.1-attnQ8-DFlash-Q4-GGUF quantized GGUF — — openmdw-1.1 coolthor 4 2.4k
mlx-community/Laguna-S-2.1-oQ3e quantized — 118B XXL openmdw-1.1 mlx-community 4 2.4k
mlx-community/Laguna-S-2.1-oQ5e quantized — 118B XXL openmdw-1.1 mlx-community 4 2.3k
mlx-community/Laguna-S-2.1-oQ4e-fast quantized — 118B XXL openmdw-1.1 mlx-community 3 2.3k
TensorFold/Laguna-S-2.1-MLX-6bit quantized MLX 118B XXL openmdw-1.1 TensorFold 2 2.1k
llmfan46/Laguna-S-2.1-Uncensored-Heretic finetune — 118B XXL openmdw-1.1 llmfan46 0 2.1k
ggml-org/Laguna-S-2.1-GGUF quantized GGUF — — openmdw-1.1 ggml-org 2 2k
xautonomics/Laguna-S-2.1-UD-IQ4_NL.LAYR.GGUF quantized GGUF — — other xautonomics 0 2k
jabbatheduck/Laguna-S-2.1-Uncensored-GGUF quantized GGUF — — openmdw-1.1 jabbatheduck 3 1.9k
pipenetwork/Laguna-S-2.1-MLX-6bit quantized MLX 118B XXL openmdw-1.1 pipenetwork 2 1.6k
unsloth/Laguna-S-2.1 finetune — 118B XXL openmdw-1.1 unsloth 7 1.3k

Repository changelog

10 commits

Repository changelog 10 commits

DateCommitSHA
2026-08-19 Update README 0f57314
2026-07-27 docs: add Ollama usage example 00af5a5
2026-07-24 tokenizer_config: inline chat_template (resolve include for GGUF conversion) b0a9fd7
2026-07-22 Recommend 7 DFlash draft tokens for llama.cpp (#11) e80da38
2026-07-21 Enable thinking by default, preserve reasoning; drop max_new_tokens cap 179ee67
2026-07-21 Mark </assistant> (token 24) as special in tokenizer.json 88796b9
2026-07-21 Mark </assistant> (token 24) as special to match internal serving 6548c10
2026-07-21 Fix chat template: preserve reasoning across turns (llama.cpp --reasoning-preserve) fac825e
2026-07-21 Fix eval task ID (#2) a50e85e
2026-07-21 Laguna S 2.1 initial release e4daa58

All providers

2 providers · 2 endpoints

Where to run 2 providers · 2 endpoints

Hosted endpoints observed for Laguna-S-2.1, sorted by input price (USD per million tokens). Each row names its observer and date; prices change daily, so check the source before quoting. Cheapest observed input price: $0.00/M.

ProviderModel refContextQuant$/M in$/M outtok/sTTFTToolsStatusObserved
Poolside
openrouter
 poolside/laguna-s-2.1 1M fp4 $0.00 $0.00 — — yes live 2026-10-24
Featherless
hf
 poolside/Laguna-S-2.1 — — — — — — — live 2026-10-25

Articles and coverage

21 pieces

Ranked by source signals (HN, Lobsters, authority domains, official posts) and recency. Summaries are generated from a stored copy of the article text; every sentence cites a span of that text, and nothing is shown that failed the citation check.

Analysis8

Laguna S 2.1
poolside.ai2026-07-21HN front pageHN 416 pts · 88 comments3,843 words
Poolside releases Laguna S 2.1, an agentic coding model with 118B total parameters and 8B activated parameters per token that launched in under nine weeks. It supports a context window of up to 1M tokens, scores 70.2% on Terminal-Bench 2.1, and has a knowledge cutoff date in November 2025. Full evaluation trajectories are available for every trial in the final evaluation set.written, reviewed, cited
  • Laguna S 2.1 scored 40.4% in our agent harness, pool, not mini-swe-agent which DeepSWE's leaderboard uses.
Laguna XS.2 and M.1
poolside.ai2026-04-28HN front pageabout the Laguna familyHN 101 pts · 50 comments3,078 words
Poolside AI releases Laguna XS.2 and Laguna M.1, agentic coding models now served at 256K context lengths with stronger scores on SWE-bench Verified and Terminal-Bench 2.0. Laguna XS.2 is a 33B total parameter MoE with 3B activated parameters, trained on 30T tokens using NVIDIA Hopper GPUs, reaching 44.5% on SWE-Bench Pro and 30.1% on Terminal-Bench 2.0. The model is available under an Apache 2.0 license and free via API and OpenRouter for a limited time.written, reviewed, cited
  • Benchmarking used a maximum of 500 steps and sandboxed execution using 8 GB RAM/2 CPUs for most tasks.
  • The Laguna series uses approx. 4.4T+ synthetic tokens to complement natural web data throughout pre-training.
  • Laguna XS.2 is supported in NVIDIA TensorRT-LLM on Day 1 with an NVFP4 version for Blackwell architecture.
Laguna XS 2.1 33B on a RTX 3090: 296 tok/s peak, 152 tok/s at 256K context
lucebox.com2026-07-11about the Laguna familyHN 20 pts · 1 comments1,373 words
Laguna XS 2.1 33B runs on a single RTX 3090 with speculative decoding to achieve 296 tok/s peak speed and maintain 152 tok/s at 256K context. The system uses KVFlash paging with an 8K pool and ring caches to handle contexts that do not physically fit in 24 GB of VRAM while keeping output quality lossless. Prefill processes 256K tokens in 67 seconds at ~3,500 tok/s using a Qwen3-0.6B drafter for scoring KV chunks.written, reviewed, cited
  • The full KV cache for 256K tokens requires approximately 20 GB, exceeding the 24 GB VRAM limit of the RTX 3090.
  • Sliding-window layers skip the pager entirely by attending a 2K position ring instead of the whole KVFlash pool.
  • The prefill drafter compresses prompts and scores KV chunk residency, falling back to recency-only LRU without it.
Running Laguna S 2.1 locally on Apple Silicon: 52 tok/s with 38.5 GB peak memory
github.com2026-07-22authorityHN 3 pts · 1 comments2,890 words
A new benchmark harness evaluates Laguna S 2.1 quantizations on Apple Silicon, finding that pipenetwork/Laguna-S-2.1-MLX-2bit leads on the M5 Max. It achieved a perfect score passing all 38 assertions and 63.86 tok/s generation throughput while using 39.31 GB peak memory. The model reached 68.49 tok/s in the fixed 256-token decode on macOS 27.0 with MLX 0.32.0.written, reviewed, cited
  • The official NVFP4 MLX build was functional but much slower in this runtime.
  • Vcruz's IQ1_S lowers the hardware floor to a 22.13 GiB payload but failed every generation assertion and 17 of 19 agentic assertions.
Agentic AI at Two Different Scales: Nanbeige4.2-3B and Laguna S2.1
kaitchup.substack.com2026-07-31HN 2 pts · 0 comments1,362 words
Nanbeige4.2-3B and Laguna S 2.1 target agentic workloads at different scales. Nanbeige4.2-3B uses a Looped Transformer with 22 physical decoder layers executed twice to achieve 44 layer executions without storing 44 independent layers, fitting on a 16GB GPU but consuming roughly 44 GiB of KV-cache memory at 256K context. Laguna S 2.1 is a 118-billion-parameter Mixture-of-Experts model activating approximately 8 billion parameters per token, featuring 48 transformer layers and 256 routed experts, requiring vLLM 0.25.0 and a 96 GB GPU to exploit its 256K context length.written, reviewed, cited
  • Nanbeige's default configuration does not enable loop-level KV sharing, potentially requiring separate key/value entries for each pass.
  • Retaining reasoning traces in conversation history risks reaching the model's maximum context length on the very next turn.
Laguna XS.2 and Laguna M.1 by Poolside
poolside.ai2026-04-28about the Laguna familyHN 5 pts · 0 comments999 words
evidence/reports/laguna-s21-strix-halo-vulkan-benchmark.md
github.comauthority7,575 words
Blackwellboy/laguna-s21-lab
github.comauthority6,067 words

Community2

Anybody tried Laguna S 2.1 (by Poolside)?
news.ycombinator.com2026-08-06HN 7 pts · 2 comments113 words
Laguna S 2.1:118B-a9B better than Qwen3.5:122B-a10B? So far, yes
news.ycombinator.com2026-08-06HN 4 pts · 2 comments157 words
This American (San Francisco) model, Laguna S 2.1:118B-a9B, outperforms Qwen3.5:122B-a10B and Sonnet in current benchmarking tests. It demonstrates better personality and natural creativity when designing UIs. Benchmarks need to include a minimum of hardware requirements, which are currently missing.written, reviewed, cited
All coverage tracked for this model (21) — append-only, nothing is ever removed
TitleSourcePublishedFirst seenSignals
Laguna S 2.1: Our Benchmarks, VRAM and Which Version to Getmodeltested.com—2026-10-24exa
evidence/reports/laguna-s21-strix-halo-vulkan-benchmark.mdgithub.com—2026-10-24exa
Laguna S 2.1 · LLM Architectures Explainedllm-architectures-explained.vercel.app—2026-10-24exa
Blackwellboy/laguna-s21-labgithub.com—2026-10-24exa
Inside Poolside Laguna S 2.1: The 118B-A8B Open-Weight MoE Built for 1M-Context Agentic Coding | Lobhari Technologieslobhari.in2026-08-202026-10-24exa
Laguna S 2.1:118B-a9B better than Qwen3.5:122B-a10B? So far, yesnews.ycombinator.com2026-08-062026-10-07hn
Anybody tried Laguna S 2.1 (by Poolside)?news.ycombinator.com2026-08-062026-10-07hn
Agentic AI at Two Different Scales: Nanbeige4.2-3B and Laguna S2.1kaitchup.substack.com2026-07-312026-10-07hn
Running Laguna S 2.1 locally on Apple Silicon: 52 tok/s with 38.5 GB peak memorygithub.com2026-07-222026-10-07hn
Laguna S 2.1: Don't Let It Think - peasterpeaster.io2026-07-222026-10-24exa
What Is Laguna S 2.1? 118B-A8B Open Coding MoEkie.ai2026-07-222026-10-24exa
Laguna S 2.1 Coding Model Tested for Agent Workthevibefather.com2026-07-222026-10-24exa
Laguna S 2.1poolside.ai2026-07-212026-10-07hn, exa
Laguna XS 2.1 33B on a RTX 3090: 296 tok/s peak, 152 tok/s at 256K contextlucebox.com2026-07-112026-10-07hn
Laguna M.1/XS.2 Technical Report [pdf]poolside.ai2026-05-262026-10-07hn
Laguna XS.2: A Coding Model Built for Closed Environments, Now Open Weightsfirethering.com2026-04-292026-10-07hn
Laguna XS.2 and Laguna M.1 by Poolsidepoolside.ai2026-04-282026-10-07hn
Laguna XS.2 and M.1poolside.ai2026-04-282026-10-07hn
Tesla is attempting lap record with a Model S Plaid+ at Laguna Seca Racewaytwitter.com2021-05-172026-10-07hn
Watch Tesla's Record-Breaking Laguna Seca Lapengadget.com2019-09-122026-10-07hn
Tesla Model S at Refuel 2013 (Laguna Seca)youtube.com2013-09-192026-10-07hn

Similar models

3 models

Similar models

  • gpt-oss-120b · 117B
    same size band (XXL), also MoE, same task (text-generation), more providers, permissive license
  • GLM-4.5-Air · 110B
    same size band (XXL), also MoE, same task (text-generation), more providers, permissive license
  • Ling-3.0-flash · 127B
    same size band (XXL), also MoE, same task (text-generation), more providers, permissive license

Sources

4 sources

Sources

Hub snapshot: 2026-10-25 · HF API: 2026-10-24 21:38 UTC · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:45 UTC · Summary: 2026-10-07 15:08 UTC · Last verified 2026-10-25

Follow Laguna-S-2.1

Watch this model

Email alerts are not open yet; the feeds above are the same data. See feeds for the item schema.