Listed together in guide “Open-source AI models”
Google's family of lightweight open-weight models derived from the same research as Gemini, currently in the Gemma 4 generation.
Open (open weights/source) Gemma Terms of Use
Use cases
- Running capable models locally on consumer hardware
- On-device and edge AI applications
- Fine-tuning small models for focused tasks
Pricing: Gemma weights are free to download under the Gemma license; hosted inference is available through Google and third parties at usage prices.
Sources
This record was last reviewed on 2026-09-10.
Listed together in guide “Open-source AI models”
Fully open large language model family from Ai2, released with its training data, code, and weights.
Open (open weights/source) Apache-2.0
Use cases
- Academic research on LLM training dynamics
- Auditing model behavior against known training data
- Building on a transparent, commercially usable base
Pricing: OLMo models, data, and code are entirely free under Apache 2.0; users bear only their own compute costs.
Sources
This record was last reviewed on 2026-10-10.
Related: shares the task Running models yourself
Open-weight bilingual LLM family from Beijing-based Baichuan Intelligence, now focused on medical AI with its Baichuan-M reasoning models.
Open (open weights/source) Baichuan Community License
Use cases
- Building Chinese-language NLP applications on open weights
- Research on bilingual large language models
- Medical AI assistants based on the Baichuan-M series
Pricing: Open-weight Baichuan models are free to download for research, with licensing terms for commercial use; hosted APIs are billed by usage.
Sources
This record was last reviewed on 2026-09-10.
Related: shares the task Running models yourself
Cohere's open-weight LLM family optimized for retrieval-augmented generation and tool use; succeeded by Command A as Cohere's flagship.
Open (open weights/source) CC-BY-NC-4.0
Use cases
- Enterprise RAG assistants grounded in company documents
- Building agents that call external tools and APIs
- Research on open-weight instruction-tuned models
Pricing: Model weights are free for non-commercial research under a CC-BY-NC license; commercial access is via Cohere's usage-priced API and enterprise deployments.
Pricing ↗ · Documentation ↗ · Sources
This record was last reviewed on 2026-10-09.
Related: shares the task Running models yourself
Open-weight mixture-of-experts LLM released by Databricks in 2024; retired from Databricks serving in 2025, though the weights remain available.
Open (open weights/source) Databricks Open Model License
Use cases
- Research on mixture-of-experts model architectures
- Self-hosted LLM deployments on enterprise infrastructure
- Fine-tuning experiments on a large open base model
Pricing: DBRX weights are free to download under the Databricks Open Model License; running the model requires the user's own infrastructure, and Databricks no longer serves it as a hosted model.
Sources
This record was last reviewed on 2026-09-10.
Related: shares the task Running models yourself
Open-weight LLM family and chat assistant from Hangzhou-based DeepSeek, known for strong reasoning models.
Open (open weights/source) MIT (DeepSeek-V3 repository code); DeepSeek Model License (DeepSeek-V3 Base and Chat weights)
Use cases
- Math, coding, and analytical tasks that need explicit reasoning
- Self-hosting open-weight models for privacy or cost control
- Cost-sensitive API integrations
Pricing: The chat assistant is free; the developer API is usage-priced, and open-weight models can be self-hosted without license fees.
Pricing ↗ · Sources
This record was last reviewed on 2026-10-13.
Related: shares the task Running models yourself
Family of open-weight large language models developed by the Technology Innovation Institute (TII).
Open (open weights/source) TII Falcon License
Use cases
- Self-hosted chatbots and text generation
- Research on large-scale open models
- Fine-tuning for domain-specific applications
Pricing: Falcon models are free to download and use commercially under permissive TII licenses; costs come only from the compute used to run them.
Sources
This record was last reviewed on 2026-09-10.
Related: shares the task Running models yourself
Family of open-weight large language models from Zhipu AI, spanning general, reasoning, and coding-focused variants.
Open (open weights/source) MIT
Use cases
- Self-hosting a capable open LLM for products
- Agentic coding via GLM coding plans and APIs
- Research on bilingual Chinese-English models
Pricing: GLM open weights are free under permissive licenses; Zhipu's hosted APIs and coding subscription plans are paid by usage or monthly fee.
Sources
This record was last reviewed on 2026-09-10.