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.

950 downloadable models

Open sourceWeights and code published under an OSI-approved licence that allows commercial use and derivatives.

1

Open weightsWeights downloadable under a permissive or Creative Commons licence allowing commercial use; code or data may be missing.

869

Restricted weightsWeights downloadable, but the licence restricts commercial use, hosting, derivatives or field of use (community, research and RAIL licences).

80

Downloadable totalUniverse = 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.

950

New in 30 daysDownloadable models with a release date inside the last 30 days (release_date claim; models without one are not counted).

269

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 (275) · Other (unclassified licence) (115) · MIT License (92) · Creative Commons Attribution-NonCommercial 4.0 (22) · Apple Sample Code / ML Research License (15) · Llama 3.2 Community License (12)

Category · this pageCommercial useRedistributionDerivativesHosting
Open source n=0————
Open weights n=3711026110261102611026
restricted-weights n=131300130013000130

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

Downloadable models

950 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
LFM2.5-350MOpen weightsLiquid AIOther354.5M—31 Mar 2026—0✓ fits0.7 GB4bit✓ fits0.9 GB8bit
LFM2.5-350M-BaseOpen weightsLiquid AI—unclassified————0——
LFM2.5-8B-A1B-BaseOpen weightsLiquid AI—unclassified————0——
LFM2.5-8B-A1B-DSparkOpen weightsLiquid AI—unclassified————0——
LFM2.5-Audio-1.5BOpen weightsLiquid AI—unclassified————0——
LFM2.5-VL-3BOpen weightsLiquid AIOther3.12B—11 Aug 2026—0✓ fits2.3 GB4bit✓ fits4.1 GB8bit
LFM2.5-VL-450MOpen weightsLiquid AIOther448.7M—8 Apr 2026—0✓ fits0.8 GB4bit✓ fits1.0 GB8bit
Laguna S 2.1Open weightsPoolside · text—unclassified—1.05M21 Jul 2026—1from $0.18 out——
Laguna XS 2.1Open weightsPoolside · text—unclassified—262.1K2 Jul 2026—1from $0.12 out——
Leanstral 1.5Open weightsMistral AIApache-2.0—256K——1——
Llama 3 8B LunarisOpen weightsSao10K · text—unclassified—8.19K13 Aug 2024—1from $0.05 out——
Llama 3.1 70B Instructrestricted-weightsMeta AI · textLlama-3.1-Community70.5B131.1K23 Jul 2024—1from $0.72 out✓ fits41.1 GB4bit✓ fits81.6 GB8bit
Llama 3.1 8B Instructrestricted-weightsMeta AI · textLlama-3.1-Community8.03B131.1K23 Jul 2024—1from $0.08 out✓ fits5.1 GB4bit✓ fits9.7 GB8bit
Llama 3.1 Euryale 70B v2.2Open weightsSao10K · text—unclassified—131.1K28 Aug 2024—1from $0.85 out——
Llama 3.2 1B Instructrestricted-weightsMeta AI · textLlama-3.2-Community1.24B60K18 Sept 2024—1from $0.201 out✓ fits1.2 GB4bit✓ fits1.9 GB8bit
Llama 3.2 3B Instructrestricted-weightsMeta AI · textLlama-3.2-Community3.21B131.1K18 Sept 2024—1from $0.33 out✓ fits2.4 GB4bit✓ fits4.2 GB8bit
Llama 3.3 70B Instructrestricted-weightsMeta AI · textLlama-3.3-Community70.5B131.1K26 Nov 2024—1from $0.32 out✓ fits41.1 GB4bit✓ fits81.6 GB8bit
Llama 3.3 Euryale 70BOpen weightsSao10K · text—unclassified—131.1K18 Dec 2024—1from $0.75 out——
Llama Guard 4 12BOpen weightsMeta AI · image, text—unclassified—163.8K30 Apr 2025—1from $0.18 out——
Llama-2-13b-chat-hfrestricted-weightsMeta AILlama-2-Community13B—13 Jul 2023—0✓ fits8.0 GB4bit✓ fits15.5 GB8bit
Llama-2-7b-chat-hfrestricted-weightsMeta AILlama-2-Community6.74B—13 Jul 2023—0✓ fits4.4 GB4bit✓ fits8.3 GB8bit
Llama-2-7b-hfrestricted-weightsMeta AILlama-2-Community6.74B—13 Jul 2023—0✓ fits4.4 GB4bit✓ fits8.3 GB8bit
Llama-3.2-11B-Vision-Instructrestricted-weightsMeta AILlama-3.2-Community10.7B—18 Sept 2024—0✓ fits6.6 GB4bit✓ fits12.8 GB8bit
Llama-3.2-90B-Vision-Instructrestricted-weightsMeta AILlama-3.2-Community88.6B—19 Sept 2024—0✓ fits51.4 GB4bit✓ fits102.4 GB8bit
Llama-4-Scout-17B-16EOpen weightsMeta AI—unclassified————0——
M1-3BOpen weightsTogether AIMIT3.45B—2 May 2025—0✓ fits2.5 GB4bit✓ fits4.5 GB8bit
Meta-Llama-3-70Brestricted-weightsMeta AILlama-3-Community70.5B—17 Apr 2024—0✓ fits41.1 GB4bit✓ fits81.6 GB8bit
Meta-Llama-3-8Brestricted-weightsMeta AILlama-3-Community8.03B—17 Apr 2024—0✓ fits5.1 GB4bit✓ fits9.7 GB8bit
Meta-Llama-3-8B-Instructrestricted-weightsMeta AILlama-3-Community8.03B—17 Apr 2024—0✓ fits5.1 GB4bit✓ fits9.7 GB8bit
Meta-Llama-3.1-70BOpen weightsMeta AI—unclassified————0——
Meta-Llama-3.1-8BOpen weightsMeta AI—unclassified————0——
Meta-Llama-3.2-3BOpen weightsMeta AI—unclassified————0——
MiniMax-01Open weightsMiniMax · image, text—unclassified456.1B1M12 Jan 2025—1from $1.1 out✗ too large262.8 GB4bit✗ too large525.0 GB8bit
MiniMax-H3Open weightsMiniMaxOther33.1B—28 Jul 2026—0✓ fits19.5 GB4bit✓ fits38.6 GB8bit
MiniMax-M1-40k-hfOpen weightsMiniMaxApache-2.0456.1B—1 Jul 2025—0✗ too large262.8 GB4bit✗ too large525.0 GB8bit
MiniMax-M1-80k-hfOpen weightsMiniMaxApache-2.0456.1B—1 Jul 2025—0✗ too large262.8 GB4bit✗ too large525.0 GB8bit
MiniMax-Music3Open weightsMiniMax—unclassified2.43B—7 Aug 2026—0✓ fits1.9 GB4bit✓ fits3.3 GB8bit
MiniMax-Text-01-hfOpen weightsMiniMaxOther456.1B—3 Jun 2025—0✗ too large262.8 GB4bit✗ too large525.0 GB8bit
MiniMax-VL-01Open weightsMiniMax—unclassified456.4B—12 Jan 2025—0✗ too large262.9 GB4bit✗ too large525.4 GB8bit
Ministral 3 14B 2512Open weightsMistral AI · image, textApache-2.013.9B262.1K31 Oct 2025—1from $0.20 out✓ fits8.5 GB4bit✓ fits16.5 GB8bit
Ministral 3 3B 2512Open weightsMistral AI · image, textApache-2.03.85B131.1K31 Oct 2025—1from $0.10 out✓ fits2.7 GB4bit✓ fits4.9 GB8bit
Ministral 3 8B 2512Open weightsMistral AI · image, text—unclassified—262.1K2 Dec 2025—1from $0.075 out——
Ministral-3-14B-Base-2512Open weightsMistral AI—unclassified————0——
Ministral-3-14B-Reasoning-2512Open weightsMistral AIApache-2.013.9B—31 Oct 2025—0✓ fits8.5 GB4bit✓ fits16.5 GB8bit
Ministral-3-3B-Base-2512Open weightsMistral AI—unclassified————0——
Ministral-3-3B-Reasoning-2512Open weightsMistral AIApache-2.04.25B—31 Oct 2025—0✓ fits3.0 GB4bit✓ fits5.4 GB8bit
Ministral-8B-Instruct-2410Open weightsMistral AIOther8.02B—15 Oct 2024—0✓ fits5.1 GB4bit✓ fits9.7 GB8bit
Mistral NemoOpen weightsMistral AI · textApache-2.012.3B131.1K17 Jul 2024—1from $0.03 out✓ fits7.5 GB4bit✓ fits14.6 GB8bit
Mistral Small 3.1 24BOpen weightsMistral AI · image, textApache-2.024B128K11 Mar 2025—1from $0.555 out✓ fits14.3 GB4bit✓ fits28.1 GB8bit
Mistral Small 3.2 24BOpen weightsMistral AI · image, textApache-2.024B256K19 Jun 2025—1from $0.20 out✓ fits14.3 GB4bit✓ fits28.1 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