Skip to content
AI Atlas
Open model frontier

Open models

Every canonical model whose weights can be downloaded, described by measurable properties — licence permissions, parameters, context, release, benchmark ranks, providers, estimated local fit. Openness is a set of observed dimensions here, never a score.

889 downloadable models

Openness explorer

Categories and licence permissions

Left: the downloadable universe by openness category. Right: what the licences on this page allow, counted per category — allowed · restricted · unknown.

Top licences: Apache License 2.0 (311) · MIT License (125) · Other (unclassified licence) (83) · Creative Commons Attribution-NonCommercial 4.0 (24) · Apple Sample Code / ML Research License (15) · Llama 3.2 Community License (12)

Category · this pageCommercial useRedistributionDerivativesHosting
Open source n=0
Open weights n=5029021290212902129021
restricted-weights n=0

Counts read: allowed · restricted · unknown / unclassified — over the 50 models on this page, from each licence's stated terms (ontology).

Downloadable models

889 models · sorted by best benchmark rank

Permissions, in order:commercial use· redistribution· derivatives· hosting allowed restricted unknown

ModelLicence · permissionsParamsContextReleaseBest resultsProvidersFit 64 GB @4bit · 128 GB @8bit Estimated
Kimi-K2-BaseOpen weightsMoonshot AI · textOther1.03T3 Jul 20250✗ too large590.7 GB4bit✗ too large1180.9 GB8bit
Kimi-Linear-48B-A3B-BaseOpen weightsMoonshot AI · textMIT49.1B3B active30 Oct 20250✓ fits28.7 GB4bit✓ fits57.0 GB8bit
Kimi-VL-A3B-InstructOpen weightsMoonshot AI · image, textMIT16.4B3B active9 Apr 20250✓ fits9.9 GB4bit✓ fits19.4 GB8bit
Kimi-VL-A3B-ThinkingOpen weightsMoonshot AI · image, textMIT16.4B3B active9 Apr 20250✓ fits9.9 GB4bit✓ fits19.4 GB8bit
Kimi-VL-A3B-Thinking-2506Open weightsMoonshot AI · image, textMIT16.4B3B active21 Jun 20250✓ fits9.9 GB4bit✓ fits19.4 GB8bit
KwaiCoder-23B-A4B-v1Open weightsKwaipilot · textMIT23.3B4B active22 Jan 20250✓ fits13.9 GB4bit✓ fits27.2 GB8bit
KwaiCoder-AutoThink-previewOpen weightsKwaipilotOther40.6B2 Jun 20250✓ fits23.8 GB4bit✓ fits47.1 GB8bit
KwaiCoder-DS-V2-Lite-BaseOpen weightsKwaipilot · textMIT15.7B2 Dec 20240✓ fits9.5 GB4bit✓ fits18.6 GB8bit
LFM2.5-1.2B-BaseOpen weightsLiquid AIunclassified0
LFM2.5-2.6B-BaseOpen weightsLiquid AIunclassified0
LFM2.5-2.6B-DSparkOpen weightsLiquid AIunclassified0
LFM2.5-230M-BaseOpen weightsLiquid AIunclassified0
LFM2.5-350M-BaseOpen weightsLiquid AIunclassified0
LFM2.5-8B-A1B-BaseOpen weightsLiquid AIunclassified0
LFM2.5-8B-A1B-DSparkOpen weightsLiquid AIunclassified0
LFM2.5-Audio-1.5BOpen weightsLiquid AIunclassified0
LLaDA2.0-UniOpen weightsinclusionAIApache-2.016.3B22 Apr 20260✓ fits9.9 GB4bit✓ fits19.3 GB8bit
LLaDA2.0-miniOpen weightsinclusionAI · textApache-2.016.3B25 Nov 20250✓ fits9.8 GB4bit✓ fits19.2 GB8bit
LLaDA2.1-miniOpen weightsinclusionAI · textApache-2.016.3B9 Feb 20260✓ fits9.8 GB4bit✓ fits19.2 GB8bit
Laguna S 2.1Open weightsPoolside · textunclassified1.05M21 Jul 20261from $0.18 out
Laguna XS 2.1Open weightsPoolside · textunclassified262.1K2 Jul 20261from $0.12 out
Leanstral 1.5Open weightsMistral AIApache-2.0256K1
Ling-3.0-flash-dsparkOpen weightsinclusionAI · textOther1.36B9 Aug 20260✓ fits1.3 GB4bit✓ fits2.1 GB8bit
Ling-lite-1.5Open weightsinclusionAI · textMIT16.8B11 May 20250✓ fits10.2 GB4bit✓ fits19.8 GB8bit
Llama 3 8B LunarisOpen weightsSao10K · textunclassified8.19K13 Aug 20241from $0.05 out
Llama 3.1 Euryale 70B v2.2Open weightsSao10K · textunclassified131.1K28 Aug 20241from $0.85 out
Llama 3.3 Euryale 70BOpen weightsSao10K · textunclassified131.1K18 Dec 20241from $0.75 out
Llama Guard 4 12BOpen weightsMeta AI · image, textunclassified163.8K30 Apr 20251from $0.18 out
Llama-4-Scout-17B-16EOpen weightsMeta AIunclassified0
M1-3BOpen weightsTogether AI · textMIT3.45B2 May 20250✓ fits2.5 GB4bit✓ fits4.5 GB8bit
Mamba2-primed-HQwen3-8B-InstructOpen weightsAmazon Web Services · textApache-2.08.5B31 Mar 20260✓ fits5.4 GB4bit✓ fits10.3 GB8bit
Meta-Llama-3.1-70BOpen weightsMeta AIunclassified0
Meta-Llama-3.1-8BOpen weightsMeta AIunclassified0
Meta-Llama-3.2-3BOpen weightsMeta AIunclassified0
MiMo-7B-BaseOpen weightsXiaomi · textMIT7.83B29 Apr 20250✓ fits5.0 GB4bit✓ fits9.5 GB8bit
MiMo-7B-RLOpen weightsXiaomi · textMIT7.83B29 Apr 20250✓ fits5.0 GB4bit✓ fits9.5 GB8bit
MiMo-7B-RL-ZeroOpen weightsXiaomi · textMIT7.83B29 Apr 20250✓ fits5.0 GB4bit✓ fits9.5 GB8bit
MiMo-7B-SFTOpen weightsXiaomi · textMIT7.83B29 Apr 20250✓ fits5.0 GB4bit✓ fits9.5 GB8bit
MiMo-Audio-7B-BaseOpen weightsXiaomiMIT8.02B18 Sept 20250✓ fits5.1 GB4bit✓ fits9.7 GB8bit
MiMo-Audio-7B-InstructOpen weightsXiaomiMIT8.02B18 Sept 20250✓ fits5.1 GB4bit✓ fits9.7 GB8bit
MiMo-Audio-TokenizerOpen weightsXiaomiMIT1.3B18 Sept 20250✓ fits1.3 GB4bit✓ fits2.0 GB8bit
MiMo-Embodied-7BOpen weightsXiaomi · image, textMIT8.31B19 Nov 20250✓ fits5.3 GB4bit✓ fits10.1 GB8bit
MiMo-V2-Flash-BaseOpen weightsXiaomi · textMIT309.8B16 Dec 20250✗ too large178.6 GB4bit✗ too large356.8 GB8bit
MiMo-V2.5-ASROpen weightsXiaomi · audio, textMIT7.62B23 Apr 20260✓ fits4.9 GB4bit✓ fits9.3 GB8bit
MiMo-V2.5-BaseOpen weightsXiaomi · textMIT310.8B27 Apr 20260✗ too large179.2 GB4bit✗ too large357.9 GB8bit
MiMo-V2.5-Pro-BaseOpen weightsXiaomi · textMIT1.02T27 Apr 20260✗ too large588.9 GB4bit✗ too large1177.2 GB8bit
MiMo-VL-7B-RLOpen weightsXiaomi · image, textMIT8.31B30 May 20250✓ fits5.3 GB4bit✓ fits10.1 GB8bit
MiMo-VL-7B-RL-2508Open weightsXiaomi · image, textMIT8.31B7 Aug 20250✓ fits5.3 GB4bit✓ fits10.1 GB8bit
MiMo-VL-7B-SFTOpen weightsXiaomi · image, textMIT8.31B30 May 20250✓ fits5.3 GB4bit✓ fits10.1 GB8bit
MiMo-VL-7B-SFT-2508Open weightsXiaomi · image, textMIT8.31B7 Aug 20250✓ fits5.3 GB4bit✓ fits10.1 GB8bit

Best results = rank inside each benchmark's primary comparability group; several dimensions are shown side by side and never combined. Fit columns are estimates (64 GB device at 4-bit, 128 GB at 8-bit, 8K context) — method on /methodology. unclassified = the raw licence label is not yet mapped in the ontology.

MethodUniverse = canonical models whose weights can be downloaded (open-weights, open-source, restricted-weights). hardware_fit values are ESTIMATES (see /methodology); best_results are ranks inside each benchmark's primary comparability group — no composite score. /methodology