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=5032018320183201832018
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
Qwen3 VL 235B A22B ThinkingOpen weightsQwen · image, textunclassified131.1K23 Sept 20251from $4 out
Qwen3 VL 30B A3B ThinkingOpen weightsQwen · image, textunclassified262.1K6 Oct 20251from $2.4 out
Qwen3 VL 8B ThinkingOpen weightsQwen · image, textunclassified131.1K14 Oct 20251from $2.1 out
Qwen3-0.6B-BaseOpen weightsQwenunclassified0
Qwen3-4B-BaseOpen weightsQwenunclassified0
Qwen3-4B-Instruct-2507Open weightsQwenunclassified0
Qwen3-8B-BaseOpen weightsQwenunclassified0
Qwen3-Embedding-0.6BOpen weightsQwen · embedding, textApache-2.0595.8M3 Jun 20250✓ fits0.8 GB4bit✓ fits1.2 GB8bit
Qwen3-VL-2B-InstructOpen weightsQwenunclassified0
Qwen3.5-4B-BaseOpen weightsQwenunclassified0
Qwen3.5-9B-BaseOpen weightsQwenunclassified0
Qwen3GuardOpen weightsQwen Team · textunclassified23 Sept 20250
ReMM SLERP 13BOpen weightsundi95 · textunclassified6.14K22 Jul 20231from $0.65 out
RedPajama-INCITE-7B-BaseOpen weightsTogether AI · textApache-2.07B4 May 20230✓ fits4.5 GB4bit✓ fits8.6 GB8bit
RedPajama-INCITE-7B-ChatOpen weightsTogether AI · textApache-2.07B4 May 20230✓ fits4.5 GB4bit✓ fits8.6 GB8bit
RedPajama-INCITE-7B-InstructOpen weightsTogether AI · textApache-2.07B5 May 20230✓ fits4.5 GB4bit✓ fits8.6 GB8bit
RedPajama-INCITE-Base-3B-v1Open weightsTogether AI · textApache-2.03B4 May 20230✓ fits2.2 GB4bit✓ fits4.0 GB8bit
RedPajama-INCITE-Chat-3B-v1Open weightsTogether AI · textApache-2.03B5 May 20230✓ fits2.2 GB4bit✓ fits4.0 GB8bit
RedPajama-INCITE-Instruct-3B-v1Open weightsTogether AI · textApache-2.03B5 May 20230✓ fits2.2 GB4bit✓ fits4.0 GB8bit
Reka EdgeOpen weightsrekaai · image, text, videounclassified16.4K20 Mar 20261from $0.10 out
Ring-2.5-1TOpen weightsinclusionAI · textMIT1.01T10 Feb 20260✗ too large582.7 GB4bit✗ too large1164.8 GB8bit
SOLAR-10.7B-v1.0Open weightsUpstage · textApache-2.010.7B12 Dec 20230✓ fits6.7 GB4bit✓ fits12.8 GB8bit
SRPO-Qwen-32BOpen weightsKwaipilotMIT32.8B21 Apr 20250✓ fits19.3 GB4bit✓ fits38.2 GB8bit
Schematron V2 SmallOpen weightsInference.net · textunclassified128K1from $0.23 out
Schematron V2 TurboOpen weightsInference.net · textunclassified128K1from $0.15 out
Seed-Coder-8B-BaseOpen weightsByteDance · textMIT8.25B27 Apr 20250✓ fits5.3 GB4bit✓ fits10.0 GB8bit
Seed-Coder-8B-InstructOpen weightsByteDance · textMIT8.25B27 Apr 20250✓ fits5.3 GB4bit✓ fits10.0 GB8bit
Seed-Coder-8B-ReasoningOpen weightsByteDance · textMIT8.25B27 Apr 20250✓ fits5.3 GB4bit✓ fits10.0 GB8bit
Seed-OSS-36B-BaseOpen weightsByteDance · textApache-2.036.1B20 Aug 20250✓ fits21.3 GB4bit✓ fits42.1 GB8bit
Seed-X-PPO-7BOpen weightsByteDance · textOther7.51B16 Jul 20250✓ fits4.8 GB4bit✓ fits9.1 GB8bit
SeedVR-3BOpen weightsByteDance · videoApache-2.03B11 Jun 20250✓ fits2.2 GB4bit✓ fits4.0 GB8bit
SeedVR-7BOpen weightsByteDance · videoApache-2.07B11 Jun 20250✓ fits4.5 GB4bit✓ fits8.6 GB8bit
SeedVR2-3BOpen weightsByteDance · videoApache-2.03B10 Jun 20250✓ fits2.2 GB4bit✓ fits4.0 GB8bit
SeedVR2-7BOpen weightsByteDance · videoApache-2.07B11 Jun 20250✓ fits4.5 GB4bit✓ fits8.6 GB8bit
Shieldstral 1.0Open weightsMistral AIApache-2.032K0
Skyfall 36B V2Open weightsTheDrummer · textunclassified32.8K10 Mar 20251from $0.80 out
SmolLM-1.7BOpen weightsHugging Face · textApache-2.01.71B14 Jul 20240✓ fits1.5 GB4bit✓ fits2.5 GB8bit
SmolLM-1.7B-InstructOpen weightsHugging Face · textApache-2.01.71B15 Jul 20240✓ fits1.5 GB4bit✓ fits2.5 GB8bit
SmolLM-135MOpen weightsHugging Face · textApache-2.0134.5M14 Jul 20240✓ fits0.6 GB4bit✓ fits0.7 GB8bit
SmolLM-135M-InstructOpen weightsHugging Face · textApache-2.0134.5M15 Jul 20240✓ fits0.6 GB4bit✓ fits0.7 GB8bit
SmolLM-360MOpen weightsHugging Faceunclassified0
SmolLM-360M-InstructOpen weightsHugging Face · textApache-2.0361.8M15 Jul 20240✓ fits0.7 GB4bit✓ fits0.9 GB8bit
SmolLM2-1.7BOpen weightsHugging Face · textApache-2.01.71B30 Oct 20240✓ fits1.5 GB4bit✓ fits2.5 GB8bit
SmolLM2-1.7B-InstructOpen weightsHugging Face · textApache-2.01.71B31 Oct 20240✓ fits1.5 GB4bit✓ fits2.5 GB8bit
SmolLM2-135MOpen weightsHugging Face · textApache-2.0134.5M31 Oct 20240✓ fits0.6 GB4bit✓ fits0.7 GB8bit
SmolLM2-135M-InstructOpen weightsHugging Face · textApache-2.0134.5M31 Oct 20240✓ fits0.6 GB4bit✓ fits0.7 GB8bit
SmolLM2-360MOpen weightsHugging Face · textApache-2.0361.8M31 Oct 20240✓ fits0.7 GB4bit✓ fits0.9 GB8bit
SmolLM2-360M-InstructOpen weightsHugging Face · textApache-2.0361.8M31 Oct 20240✓ fits0.7 GB4bit✓ fits0.9 GB8bit
SmolLM3-3BOpen weightsHugging Face · textApache-2.03.08B8 Jul 20250✓ fits2.3 GB4bit✓ fits4.0 GB8bit
SmolLM3-3B-BaseOpen weightsHugging Face · textApache-2.03.08B19 Jun 20250✓ fits2.3 GB4bit✓ fits4.0 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