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.

870 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.

870

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

79

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.

870

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=5010040100401004010040
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

870 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
Qwen3.5-0.8BOpen weightsQwen—unclassified—262.1K—tau2-bench#125mmmu-pro#155terminal-bench#2790——
nvidia-nemotron-nano-12b-v2-vlOpen weightsNVIDIA—unclassified—128K—mmmu-pro#125terminal-bench#210tau2-bench#2420——
Ministral 3 8BOpen weightsMistral AIApache-2.0—256K—scicode#126mmmu-pro#135terminal-bench#2101from $0.15 out——
Qwen3-VL-4B-InstructOpen weightsQwenApache-2.04.44B256K11 Oct 2025mmmu-pro#126ifbench#229tau2-bench#2290✓ fits3.0 GB4bit✓ fits5.6 GB8bit
Gemma 3 12BOpen weightsGoogle · image, text—unclassified—131.1K13 Mar 2025scicode#127mmmu-pro#147ifbench#2272from $0.15 out——
falcon-h1r-7bOpen weightsTII UAE—unclassified—256K—ifbench#127humanitys-last-exam#184tau2-bench#2020——
gemma-4-E4BOpen weightsGoogle—unclassified—128K—mmmu-pro#127ifbench#173terminal-bench#1760——
Ministral 3 3BOpen weightsMistral AIApache-2.0—256K—scicode#128mmmu-pro#144tau2-bench#2231from $0.10 out——
Pixtral LargeOpen weightsMistral AI—unclassified—128K—mmmu-pro#128tau2-bench#170ifbench#2440——
GLM 4.5VOpen weightsZ.ai (Zhipu AI) · image, text—unclassified—65.5K11 Aug 2025mmmu-pro#129terminal-bench#186tau2-bench#2352from $1.8 out——
LFM2.5-2.6B (free)Open weightsLiquid AIOther2.7B128K28 Jul 2026scicode#129humanitys-last-exam#266artificial-analysis-intelligence-index#2791✓ fits2.0 GB4bit✓ fits3.6 GB8bit
GLM 4.6VOpen weightsZ.ai (Zhipu AI) · image, text, video—unclassified—131.1K8 Dec 2025mmmu-pro#131terminal-bench#148tau2-bench#1862from $0.90 out——
deepseek-r1Open weightsDeepSeekMIT—128K—gpqa-diamond#131terminal-bench#141humanitys-last-exam#1461from $2.19 out——
ernie-4-5-300b-a47bOpen weightsBaidu—unclassified—131.1K—gpqa-diamond#134terminal-bench#198ifbench#2080——
cogito-v2-1-reasoningOpen weightsDeep Cogito—unclassified—128K—terminal-bench#137ifbench#162gpqa-diamond#1710——
gemma-4-E2BOpen weightsGoogle—unclassified—128K—mmmu-pro#138ifbench#219terminal-bench#2330——
jamba-reasoning-3bOpen weightsAI21 Labs—unclassified—262K—ifbench#139terminal-bench#265tau2-bench#2680——
Seed-OSS-36B-InstructOpen weightsByteDanceApache-2.036.1B512K20 Aug 2025tau2-bench#143terminal-bench#186ifbench#1860✓ fits21.3 GB4bit✓ fits42.1 GB8bit
llama-3-2-instruct-90b-visionOpen weightsMeta AI—unclassified—128K—mmmu-pro#143humanitys-last-exam#348gpqa-diamond#3570——
Qwen3 Coder 30B A3B InstructOpen weightsQwen · text—unclassified—262.1K31 Jul 2025terminal-bench#144tau2-bench#174artificial-analysis-intelligence-index#2411from $0.28 out——
Molmo2-8BOpen weightsAllen Institute for AIApache-2.08.66B36.9K14 Dec 2025mmmu-pro#148terminal-bench#279ifbench#2920✓ fits5.5 GB4bit✓ fits10.5 GB8bit
qwen3-4b-2507-instructOpen weightsAlibaba Group—unclassified—262.1K—ifbench#149terminal-bench#210tau2-bench#2100——
sarvam-105bOpen weightsSarvam—unclassified—128K—tau2-bench#150humanitys-last-exam#182gpqa-diamond#1960——
llama-3-2-instruct-11b-visionOpen weightsMeta AI—unclassified—128K—mmmu-pro#152terminal-bench#265tau2-bench#2720——
LFM2.5-VL-1.6BOpen weightsLiquid AIOther1.6B32K5 Jan 2026mmmu-pro#153ifbench#257terminal-bench#2790✓ fits1.4 GB4bit✓ fits2.3 GB8bit
olmo-3-32b-thinkOpen weightsAllen Institute for AI—unclassified—65.5K—ifbench#153humanitys-last-exam#256terminal-bench#2560——
gemma-3n-e4bOpen weightsGoogle—unclassified—32K—mmmu-pro#154terminal-bench#240ifbench#2890——
molmo-7b-dOpen weightsAllen Institute for AI—unclassified—4.1K—mmmu-pro#156terminal-bench#279tau2-bench#3020——
phi-4-multimodalOpen weightsMicrosoft—unclassified—128K—mmmu-pro#157humanitys-last-exam#312artificial-analysis-intelligence-index#3970——
Hermes 4 405BOpen weightsNous Research · text—unclassified—131.1K26 Aug 2025terminal-bench#159humanitys-last-exam#187gpqa-diamond#2061from $3 out——
Kimi-Linear-48B-A3B-InstructOpen weightsMoonshot AIMIT49.1B3B active1M30 Oct 2025terminal-bench#159ifbench#288tau2-bench#3020✓ fits28.7 GB4bit✓ fits57.0 GB8bit
hermes-4-llama-3-1-405bOpen weightsNous Research—unclassified—128K—terminal-bench#159humanitys-last-exam#187tau2-bench#2100——
Llama 3.3 70BOpen weightsMeta AI—unclassified—128K—ifbench#160tau2-bench#210terminal-bench#2331from $1.04 out——
granite-4.1-30bOpen weightsIBMApache-2.028.9B131.1K6 Apr 2026tau2-bench#160ifbench#171terminal-bench#2400✓ fits17.1 GB4bit✓ fits33.7 GB8bit
intellect-3Open weightsPrime Intellect—unclassified—131.1K—humanitys-last-exam#160terminal-bench#172gpqa-diamond#1780——
lfm2-24b-a2bOpen weightsLiquid AI—unclassified—32.8K—ifbench#163terminal-bench#279tau2-bench#2900——
ling-1tOpen weightsinclusionAI—unclassified—128K—terminal-bench#164tau2-bench#182gpqa-diamond#2140——
ling-flash-2-0Open weightsinclusionAI—unclassified—128K—terminal-bench#164ifbench#246tau2-bench#2480——
ring-1tOpen weightsinclusionAI—unclassified—128K—gpqa-diamond#164ifbench#169humanitys-last-exam#1800——
Devstral Small 1.0Open weightsMistral AI—unclassified—256K—tau2-bench#166terminal-bench#198ifbench#2680——
Qwen2.5 72B InstructOpen weightsQwen · text—unclassified—32.8K19 Sept 2024tau2-bench#174terminal-bench#210ifbench#2261from $0.40 out——
sarvam-30bOpen weightsSarvam—unclassified—65.5K—tau2-bench#174humanitys-last-exam#234terminal-bench#2400——
LFM2.5-1.2B-InstructOpen weightsLiquid AIOther1.17B32K6 Jan 2026ifbench#175humanitys-last-exam#251terminal-bench#2790✓ fits1.2 GB4bit✓ fits1.9 GB8bit
exaone-4-0-32bOpen weightsLG AI Research—unclassified—131K—humanitys-last-exam#177gpqa-diamond#194terminal-bench#2220——
MiniMax-M1-80kOpen weightsMiniMaxApache-2.0456.1B1M13 Jun 2025tau2-bench#178ifbench#188artificial-analysis-intelligence-index#2070✗ too large262.8 GB4bit✗ too large525.0 GB8bit
ring-flash-2-0Open weightsinclusionAI—unclassified—128K—ifbench#178terminal-bench#182gpqa-diamond#2100——
Mistral Large 2.0Open weightsMistral AI · document, text—unclassified—131.1K19 Nov 2024tau2-bench#181ifbench#268gpqa-diamond#3402from $6 out——
qwen3-235b-a22b-instructOpen weightsAlibaba Group—unclassified—32.8K—humanitys-last-exam#184terminal-bench#198tau2-bench#2070——
MiniMax-M1-40kOpen weightsMiniMaxApache-2.0456.1B1M5 Jun 2025tau2-bench#186ifbench#196humanitys-last-exam#2290✗ too large262.8 GB4bit✗ too large525.0 GB8bit
qwen3-30b-a3b-instructOpen weightsAlibaba Group—unclassified—32.8K—terminal-bench#186ifbench#190tau2-bench#2170——

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