Open weights are a deployment decision, not a magic label

A frontier-scale open-weight model with a million-token context and long-horizon agent ambitions.

Licence

Kimi K3 License

Scale

2.8T total / 104B active MoE

Practical fit: Best for teams evaluating provider independence and long-context work, not casual local inference.

Deployment: The checkpoint is roughly a datacentre deployment; review the custom licence before redistribution.

An official post-trained V4-Flash checkpoint focused on coding, agents, reasoning, and tool use.

Licence

MIT

Scale

284B total / 13B active MoE

Practical fit: A compelling hosted API price-performance option with open deployment for teams that can carry the hardware.

Deployment: Only a small fraction activates per token; the full expert set still has to be served.

An open multimodal mixture-of-experts release designed for broad agent and reasoning experiments.

Licence

Apache 2.0

Scale

975B total / 41B active MoE

Practical fit: Useful for teams building an open-model evaluation lane, especially where modality breadth matters.

Deployment: Total parameter count makes quantisation, serving, and memory planning central to the decision.

A smaller Inkling checkpoint aimed at bringing much of the family’s capability into a more practical serving envelope.

Licence

Apache 2.0

Scale

276B total / 12B active MoE

Practical fit: The better first experiment for teams that want to test the family without starting with the full checkpoint.

Deployment: Smaller does not mean laptop-sized; estimate memory and throughput from the actual release artefacts.

A large sovereign-model release from LG AI Research for teams evaluating Korean and enterprise-focused open infrastructure.

Licence

Apache 2.0

Scale

750B foundation model

Practical fit: Interesting for regional language, sovereignty, and research use cases where a broad open ecosystem matters.

Deployment: Confirm the actual checkpoint, serving recipes, and hardware requirements before treating it as a production option.

A smaller policy-adaptive safety model designed to sit around a generation system rather than replace it.

Licence

Apache 2.0

Scale

3B classifier

Practical fit: Useful for moderation, routing, and safety layers where a compact model can reduce latency and cost.

Deployment: Evaluate false positives and policy fit on your own content; a classifier is not a complete safety programme.

A compact on-device agent and tool-calling model for latency-sensitive local workflows.

Licence

LFM Open License v1.0

Scale

2.6B

Practical fit: A better candidate for edge and private inference than the frontier-sized releases in this list.

Deployment: Commercial use is free below $10M annual organisational revenue under the LFM Open License v1.0; measure tool-call reliability and instruction following at your target memory and quantisation level.

An open vision-language-action model aimed at embodied and autonomous systems rather than general chat.

Licence

OpenMDW-1.1

Scale

34B total / 32B VLM backbone + action expert

Practical fit: Relevant to robotics and physical-world evaluation, not a drop-in chatbot replacement.

Deployment: Hardware, simulator, sensor, and safety validation are part of the product decision.

Meta’s open-weight multimodal agentic model distilled from Muse Spark for local consumer hardware.

Licence

Apache 2.0

Scale

~29.6B dense (incl. ~1.8B perception encoder)

Practical fit: The practical local-agent path when Muse Spark API is too heavy or closed for your deployment.

Deployment: Full BF16 needs far more than 24GB — plan on official GGUF / 4-bit packs (and mmproj for vision). Validate agentic quality on your workflows.

InclusionAI’s cost-focused Flash MoE tier for high-frequency hybrid reasoning and agent loops.

Licence

MIT

Scale

124B total / 5.1B active MoE

Practical fit: A budget open-weight option when you need MoE efficiency and can host the expert set.

Deployment: ~262K context is smaller than many frontier peers — confirm serving provider and licence before volume commit.

Official Hugging Face checkpoint for the Qwen3.8 Max-class open weights — text-only, thinking required, native 262K context.

Licence

Qwen3.8-Max License

Scale

2.4T total / 95B active MoE

Practical fit: The self-host path for Qwen3.8 flagship weights. It is not a drop-in for the hosted Qwen3.8-Max API's vision, tools, or 1M default context.

Deployment: Datacentre-class memory. YaRN can extend toward ~1,010,000 tokens. MaaS / AI Work Assistant businesses over $50M trailing 12-month revenue need a separate commercial licence.

Apache 2.0 Qwen3.8 dense VLM with native 262K context and image/video input — the practical local sibling of hosted Qwen3.8-Max.

Licence

Apache 2.0

Scale

27B dense

Practical fit: The first Qwen3.8 local target when 2.4T-A95B is too large and you can accept a dense 27B serving envelope.

Deployment: Hosted 1M-context API is coming soon. Vendor DeepSWE/SWE-bench Pro numbers used Claude Code, not the public Datacurve or swebench.com boards.