---
title: "Open weights you can actually run · Undominated.ai"
canonical: https://undominated.ai/open-weights/
description: "25 catalogue models marked openWeights=true. 11 have an independent score. Null is not open."
---

# Open weights you can actually run · Undominated.ai

> 25 catalogue models marked openWeights=true. 11 have an independent score. Null is not open.

# Open weights

Only rows with openWeights=true. Null is not a yes. Quality leads; unrated sits below, never as a win. Sizes this page does not have are not guessed.

As of Oct 6, 2026, 25 catalogue models are marked openWeights=true. 11 of them have an independent score. 5 are undominated among those rated and priced open-weight rows.

The self-host fitter skips models with no sourced size. MoE memory uses total parameters, not active. This list is the catalogue flag, not a Hugging Face scrape.

[Fit these to your hardware](/self-host/)

| Model | Quality | Input $/M | Context | Licence |
| --- | --- | --- | --- | --- |
| [Kimi K3](/models/moonshotai__kimi-k3/) | 1475.9 | $0.680 | 1M | kimi-k3 |
| [GLM 5.3](/models/z-ai__glm-5.3/) | 1471.4 | $0.200 | 1M | glm-5.3 |
| [GLM 5.2](/models/z-ai__glm-5.2/) | 1470.4 | $0.309 | 1M | mit |
| [DeepSeek V4 Pro 0423](/models/deepseek__deepseek-v4-pro/) | 1445.2 | $0.450 | 1M | — |
| [Gemma 4 31B](/models/google__gemma-4-31b-it/) | 1443.4 | $0.140 | 262K | apache-2.0 |
| [Qwen3.8 27B](/models/qwen__qwen3.8-27b/) | 1440.7 | $0.089 | 1M | apache-2.0 |
| [DeepSeek V4 Flash 0423](/models/deepseek__deepseek-v4-flash/) | 1432.1 | $0.090 | 1M | — |
| [MiniMax M3](/models/minimax__minimax-m3/) | 1432.1 | $0.280 | 1M | minimax-community |
| [DeepSeek V3.2](/models/deepseek__deepseek-v3.2/) | 1424.5 | $0.260 | 163K | mit |
| [gpt-oss-120b](/models/openai__gpt-oss-120b/) | 1365.4 | $0.030 | 131K | apache-2.0 |
| [Phi 4](/models/microsoft__phi-4/) | 1216.7 | $0.070 | 16K | mit |
| [Llama 4 Scout](/models/meta-llama__llama-4-scout/) | — | $0.100 | 1.3M | llama-4-community |
| [Ministral 3 3B 2512](/models/mistralai__ministral-3b-2512/) | — | $0.100 | 131K | apache-2.0 |
| [Ministral 3 8B 2512](/models/mistralai__ministral-8b-2512/) | — | $0.150 | 262K | apache-2.0 |
| [Mistral Small 4](/models/mistralai__mistral-small-2603/) | — | $0.150 | 262K | apache-2.0 |
| [Ministral 3 14B 2512](/models/mistralai__ministral-14b-2512/) | — | $0.200 | 262K | apache-2.0 |
| [Llama 4 Maverick](/models/meta-llama__llama-4-maverick/) | — | $0.270 | 1M | llama-4-community |
| [Command A+](/models/cohere__command-a-plus/) | — | $0.300 | 192K | apache-2.0 |
| [Muse Glimmer 30B](/models/meta__muse-glimmer-30b/) | — | $0.300 | 131K | — |
| [DeepSeek V4 Flash Vision Exp](/models/deepseek__deepseek-v4-flash-vision-exp/) | — | $0.440 | 1M | mit |
| [Mistral Large 3 2512](/models/mistralai__mistral-large-2512/) | — | $0.500 | 262K | apache-2.0 |
| [Nemotron 3 Ultra](/models/nvidia__nemotron-3-ultra-550b-a55b/) | — | $0.625 | 262K | openmdw-1.1 |
| [Kimi K2.7 Code](/models/moonshotai__kimi-k2.7-code/) | — | $0.671 | 262K | modified-mit |
| [Qwen3.8 2.4T A95B](/models/qwen__qwen3.8-2.4t-a95b/) | — | $2.00 | 1M | qwen3.8-max |
| [Command A](/models/cohere__command-a/) | — | $2.50 | 256K | cc-by-nc-4.0 |

14 have no independent score. They are listed after the ranked rows.

## Continue your investigation

 - [Estimate memory capacity](/self-host/)
- [Inspect checkpoints](/families/)
- [Compare requirements](/compare/)
- [Understand licence labels](/methodology/)