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=492047204720472047
restricted-weights n=1010100100010

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
apertus-8b-instructOpen weightsSwiss AI Initiative—unclassified—65.5K—terminal-bench#276tau2-bench#287humanitys-last-exam#3110——
gemma-3-1bOpen weightsGoogle—unclassified—32K—terminal-bench#276humanitys-last-exam#291tau2-bench#2930——
gemma-3-270mOpen weightsGoogle—unclassified—32K—terminal-bench#276tau2-bench#297ifbench#3330——
granite-3-3-8b-instructOpen weightsIBM—unclassified—128K—terminal-bench#276tau2-bench#293ifbench#3140——
granite-4-0-350mOpen weightsIBM—unclassified—32.8K—terminal-bench#276tau2-bench#279humanitys-last-exam#2810——
lfm2-8b-a1bOpen weightsLiquid AI—unclassified—32.8K—terminal-bench#276tau2-bench#293ifbench#2960——
llama-3-instruct-8bOpen weightsMeta AI—unclassified—8.19K—terminal-bench#276tau2-bench#302humanitys-last-exam#3040——
olmo-2-7bOpen weightsAllen Institute for AI—unclassified—4.1K—terminal-bench#276humanitys-last-exam#287tau2-bench#3020——
phi-3-miniOpen weightsMicrosoft—unclassified—4.1K—terminal-bench#276tau2-bench#302ifbench#3080——
phi-4-miniOpen weightsMicrosoft—unclassified—128K—terminal-bench#276tau2-bench#299ifbench#3210——
tiny-aya-globalrestricted-weightsCohere · textCC-BY-NC-4.03.35B8K13 Feb 2026terminal-bench#276humanitys-last-exam#296tau2-bench#3020✓ fits2.4 GB4bit✓ fits4.3 GB8bit
llama-3-1-nemotron-nano-4b-reasoningOpen weightsNVIDIA—unclassified—128K—tau2-bench#286humanitys-last-exam#298ifbench#3000——
deepseek-coder-v2-liteOpen weightsDeepSeek—unclassified—128K—humanitys-last-exam#289gpqa-diamond#413artificial-analysis-intelligence-index#4270——
deepseek-r1-distill-qwen-14bOpen weightsDeepSeek—unclassified—128K—artificial-analysis-intelligence-index#291ifbench#316gpqa-diamond#3370——
llama-2-chat-7bOpen weightsMeta AI—unclassified—4.1K—humanitys-last-exam#291artificial-analysis-intelligence-index#402gpqa-diamond#4540——
llama-2-chat-70bOpen weightsMeta AI—unclassified—4.1K—humanitys-last-exam#297gpqa-diamond#406artificial-analysis-intelligence-index#4300——
jamba-1-5-miniOpen weightsAI21 Labs—unclassified—256K—humanitys-last-exam#299gpqa-diamond#421artificial-analysis-intelligence-index#4400——
Mistral 7BOpen weightsMistral AI—unclassified—8.19K—tau2-bench#302ifbench#324humanitys-last-exam#3400——
QwQ-32B-PreviewOpen weightsAlibaba Group · text—unclassified32B32.8K28 Nov 2024gpqa-diamond#304artificial-analysis-intelligence-index#305humanitys-last-exam#3950✓ fits18.9 GB4bit✓ fits37.3 GB8bit
tulu3-405bOpen weightsAllen Institute for AI—unclassified—128K—gpqa-diamond#317artificial-analysis-intelligence-index#329humanitys-last-exam#4360——
Qwen2.5-Coder-7BOpen weightsQwen—unclassified—131.1K—humanitys-last-exam#318gpqa-diamond#396artificial-analysis-intelligence-index#3990——
gemma-3n-e4b-preview-0520Open weightsGoogle—unclassified—32K—humanitys-last-exam#322artificial-analysis-intelligence-index#397gpqa-diamond#4350——
command-r-03-2024Open weightsCohere · text—unclassified—128K—humanitys-last-exam#326gpqa-diamond#433artificial-analysis-intelligence-index#4520——
grok-2Open weightsxAI—unclassified—131.1K22 Aug 2025gpqa-diamond#326artificial-analysis-intelligence-index#334humanitys-last-exam#4430——
openchat-35Open weightsOpenChat—unclassified—8.19K—humanitys-last-exam#326artificial-analysis-intelligence-index#434gpqa-diamond#4510——
llama-2-chat-13bOpen weightsMeta AI—unclassified—4.1K—humanitys-last-exam#328gpqa-diamond#412artificial-analysis-intelligence-index#4300——
deepseek-r1-distill-llama-8bOpen weightsDeepSeek—unclassified—128K—ifbench#329artificial-analysis-intelligence-index#362gpqa-diamond#4210——
DeepSeek-R1-Distill-Qwen-1.5BOpen weightsDeepSeekMIT1.78B128K20 Jan 2025ifbench#332artificial-analysis-intelligence-index#414humanitys-last-exam#4420✓ fits1.5 GB4bit✓ fits2.5 GB8bit
Mixtral 8x7BOpen weightsMistral AI—unclassified—32.8K—humanitys-last-exam#332gpqa-diamond#428artificial-analysis-intelligence-index#4470——
command-r-plus-04-2024Open weightsCohere · text—unclassified—128K—humanitys-last-exam#335gpqa-diamond#410artificial-analysis-intelligence-index#4300——
qwen2.5-32b-instructOpen weightsAlibaba Group—unclassified—128K—artificial-analysis-intelligence-index#345gpqa-diamond#345humanitys-last-exam#3910——
deepseek-v2-5Open weightsDeepSeek—unclassified—128K—artificial-analysis-intelligence-index#354gpqa-diamond#3630——
jamba-1-6-miniOpen weightsAI21 Labs—unclassified—256K—humanitys-last-exam#355gpqa-diamond#424artificial-analysis-intelligence-index#4450——
qwen3-4b-instructOpen weightsAlibaba Group—unclassified—32K—artificial-analysis-intelligence-index#355gpqa-diamond#377humanitys-last-exam#4350——
jamba-1-5-largeOpen weightsAI21 Labs—unclassified—256K—gpqa-diamond#359artificial-analysis-intelligence-index#382humanitys-last-exam#3850——
deephermes-3-llama-3-1-8b-previewOpen weightsNous Research—unclassified—128K—humanitys-last-exam#360gpqa-diamond#439artificial-analysis-intelligence-index#4510——
deepseek-v2-5-sep-2024Open weightsDeepSeek—unclassified—128K—artificial-analysis-intelligence-index#3600——
mistral-smallOpen weightsMistral AI—unclassified—32.8K—humanitys-last-exam#364gpqa-diamond#381artificial-analysis-intelligence-index#3940——
r1-1776Open weightsPerplexity AI—unclassified—128K—artificial-analysis-intelligence-index#3650——
grok-1Open weightsxAIApache-2.0—8.19K17 Mar 2024artificial-analysis-intelligence-index#3730——
Hermes 3 70B InstructOpen weightsNous Research · text—unclassified—131.1K18 Aug 2024gpqa-diamond#374artificial-analysis-intelligence-index#383humanitys-last-exam#3911from $0.70 out——
qwen2-72b-instructOpen weightsAlibaba Group—unclassified—131.1K—artificial-analysis-intelligence-index#374gpqa-diamond#384humanitys-last-exam#4130——
deephermes-3-mistral-24b-previewOpen weightsNous Research—unclassified—32K—artificial-analysis-intelligence-index#378gpqa-diamond#380humanitys-last-exam#4050——
jamba-1-6-largeOpen weightsAI21 Labs—unclassified—256K—gpqa-diamond#379artificial-analysis-intelligence-index#386humanitys-last-exam#4170——
deepseek-coder-v2Open weightsDeepSeek—unclassified—128K—artificial-analysis-intelligence-index#3850——
Mixtral 8x22BOpen weightsMistral AI—unclassified—65.4K—humanitys-last-exam#390artificial-analysis-intelligence-index#401gpqa-diamond#4010——
dbrxOpen weightsDatabricks—unclassified—32.8K—gpqa-diamond#402artificial-analysis-intelligence-index#434humanitys-last-exam#4450——
qwen1.5-110b-chatOpen weightsAlibaba Group—unclassified—32K—artificial-analysis-intelligence-index#406gpqa-diamond#4300——
deepseek-v2Open weightsDeepSeek—unclassified—128K—artificial-analysis-intelligence-index#4140——
arctic-instructOpen weightsSnowflake—unclassified—4K—artificial-analysis-intelligence-index#4220——

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