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

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=1100100100100
Open weights n=4526019260192601925020
restricted-weights n=4310400400040

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
chronos-t5-smallOpen weightsAmazon Web ServicesApache-2.046.1M21 Feb 20240✓ fits0.5 GB4bit✓ fits0.6 GB8bit
chronos-t5-tinyOpen weightsAmazon Web ServicesApache-2.08.39M28 Feb 20240✓ fits0.5 GB4bit✓ fits0.5 GB8bit
clip-vit-base-patch16Open weightsOpenAIunclassified2 Mar 20220
clip-vit-base-patch32Open weightsOpenAIunclassified2 Mar 20220
clip-vit-large-patch14Open weightsOpenAIunclassified427.6M2 Mar 20220✓ fits0.7 GB4bit✓ fits1.0 GB8bit
clip-vit-large-patch14-336Open weightsOpenAIunclassified22 Apr 20220
codegen-350M-monoOpen weightsSalesforceBSD-3-Clause350M11 Apr 20220✓ fits0.7 GB4bit✓ fits0.9 GB8bit
codegen2-16B_POpen weightsSalesforceApache-2.026 Apr 20230
cohere-transcribe-03-2026Open sourceCohereApache-2.02.07B24 Mar 20260✓ fits1.7 GB4bit✓ fits2.9 GB8bit
cohere-transcribe-arabic-07-2026Open weightsCohereApache-2.02.07B18 Jun 20260✓ fits1.7 GB4bit✓ fits2.9 GB8bit
command-a-plus-05-2026Open weightsCohere · image, textunclassified128K0
command-a-vision-07-2025restricted-weightsCohere · image, textCC-BY-NC-4.0111.9B128K28 Jul 20250✗ too large64.8 GB4bit✗ too large129.2 GB8bit
ctrlOpen weightsSalesforceBSD-3-Clause2 Mar 20220
darkmatter-gpt2-small-x343Open weightsStanford CRFMunclassified2 Mar 20220
deberta-v3-baseOpen weightsMicrosoftMIT2 Mar 20220
deberta-v3-largeOpen weightsMicrosoftMIT2 Mar 20220
deberta-v3-smallOpen weightsMicrosoftMIT2 Mar 20220
deeplabv3-mobilevit-xx-smallOpen weightsAppleOther30 May 20220
deepseek-coder-6.7b-instructOpen weightsDeepSeekOther6.74B29 Oct 20230✓ fits4.4 GB4bit✓ fits8.3 GB8bit
deepseek-coder-7b-instruct-v1.5Open weightsDeepSeekOther6.91B25 Jan 20240✓ fits4.5 GB4bit✓ fits8.4 GB8bit
durin-gpt2-medium-x343Open weightsStanford CRFMunclassified2 Mar 20220
electra-base-discriminatorOpen weightsGoogleApache-2.02 Mar 20220
embeddinggemma-300mrestricted-weightsGoogleGemma-Terms302.9M17 Jul 20250✓ fits0.7 GB4bit✓ fits0.8 GB8bit
enformer-official-roughOpen weightsEleutherAICC-BY-4.01 Jun 20220
eowyn-gpt2-medium-x777Open weightsStanford CRFMunclassified2 Mar 20220
evo-1-131k-baseOpen weightsTogether AIApache-2.06.45B20 Feb 20240✓ fits4.2 GB4bit✓ fits7.9 GB8bit
evo-1-8k-baseOpen weightsTogether AIApache-2.06.45B24 Feb 20240✓ fits4.2 GB4bit✓ fits7.9 GB8bit
expanse-gpt2-small-x777Open weightsStanford CRFMunclassified2 Mar 20220
firefunction-v1Open weightsFireworks AIApache-2.046.7B16 Feb 20240✓ fits27.4 GB4bit✓ fits54.2 GB8bit
fireworks-ai/llama-3-firefunction-v2restricted-weightsFireworks AILlama-3-Community70.5B5 Jun 20240✓ fits41.1 GB4bit✓ fits81.6 GB8bit
fireworks-ai/llama-68m-testOpen weightsFireworks AIunclassified68M9 Aug 20240✓ fits0.5 GB4bit✓ fits0.6 GB8bit
fireworks-ai/mistral-7b-eagle-head-experimentalOpen weightsFireworks AIApache-2.07B30 May 20240✓ fits4.5 GB4bit✓ fits8.6 GB8bit
fireworks-ai/mixtral-8x22b-instruct-ohOpen weightsFireworks AIApache-2.0140.6B12 Apr 20240✗ too large81.4 GB4bit✗ too large162.2 GB8bit
flan-t5-baseOpen weightsGoogleApache-2.0247.6M21 Oct 20220✓ fits0.6 GB4bit✓ fits0.8 GB8bit
function-calling-v1Open weightsFireworks AIunclassified46.7B17 Feb 20240✓ fits27.4 GB4bit✓ fits54.2 GB8bit
gemma-3-12b-ptOpen weightsGoogleunclassified0
gemma-3-1b-itrestricted-weightsGoogleGemma-Terms999.9M10 Mar 20250✓ fits1.1 GB4bit✓ fits1.6 GB8bit
gemma-3-1b-ptOpen weightsGoogleunclassified0
gemma-3-4b-ptOpen weightsGoogleunclassified0
gemma-4-12B-itOpen weightsGoogleApache-2.012B23 May 20260✓ fits7.4 GB4bit✓ fits14.3 GB8bit
gemma-4-E2B-itOpen weightsGoogleApache-2.05.12B2 Mar 20260✓ fits3.4 GB4bit✓ fits6.4 GB8bit
gemma-4-E4B-itOpen weightsGoogleApache-2.08B2 Mar 20260✓ fits5.1 GB4bit✓ fits9.7 GB8bit
glm-4-9b-chatOpen weightsZ.ai (Zhipu AI)Other9.4B4 Jun 20240✓ fits5.9 GB4bit✓ fits11.3 GB8bit
gpt-j-6bOpen weightsEleutherAIApache-2.06B2 Mar 20220✓ fits4.0 GB4bit✓ fits7.4 GB8bit
gpt-neo-1.3BOpen weightsEleutherAIMIT1.37B2 Mar 20220✓ fits1.3 GB4bit✓ fits2.1 GB8bit
gpt-neo-125mOpen weightsEleutherAIMIT150.4M2 Mar 20220✓ fits0.6 GB4bit✓ fits0.7 GB8bit
gpt-neox-20bOpen weightsEleutherAIApache-2.020.7B7 Apr 20220✓ fits12.4 GB4bit✓ fits24.4 GB8bit
gpt-oss-safeguard-20bOpen weightsOpenAI · textApache-2.021.5B131.1K18 Sept 20252from $0.30 out✓ fits12.9 GB4bit✓ fits25.2 GB8bit
granite-3.0-8b-baseOpen weightsIBMunclassified0
granite-3.0-8b-instructOpen weightsIBMApache-2.08.17B2 Oct 20240✓ fits5.2 GB4bit✓ fits9.9 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