Multiple model categories
Liquid Foundation Models span text, vision-language, audio, omni, nano, and decision categories, with model availability varying by release.
Liquid Foundation Models are an open-weight and API-accessible family of text, vision-language, audio, omni, nano, and decision models for local devices, edge systems, self-managed infrastructure, and cloud applications. They help developers and organizations choose between hosted inference, local deployment, fine-tuning, and custom enterprise delivery.
Liquid Foundation Models (LFMs) are a family of foundation models from Liquid AI for text, vision-language, audio, omni, nano, and decision tasks. The product line is designed for deployment across local devices, edge systems, and cloud infrastructure rather than being limited to a single hosted application.
Users can access supported models through Liquid's API and listed distribution platforms, or download and run open weight models on their own infrastructure. Liquid also offers custom model optimization and deployment for organizations that need an on-premises, OEM, or environment-specific implementation. Model capabilities and available formats differ by individual release, so selection should be based on the relevant model's documentation.
Liquid Foundation Models span text, vision-language, audio, omni, nano, and decision categories, with model availability varying by release.
Liquid positions its models for execution on CPUs, GPUs, and NPUs across phones, laptops, vehicles, and other edge hardware, as well as in cloud environments.
Selected local distribution formats include GGUF, MLX, and ONNX, allowing supported models to run on laptops, phones, and edge boxes without an API key.
The Liquid API is described as an OpenAI-compatible endpoint. Models are also listed through Liquid Console, OpenRouter, and Vercel AI Gateway.
Open weight models can be downloaded, run, and fine-tuned. The LFM Open License does not require users to open-source modifications or derivative models.
Liquid offers custom model building and optimization for an organization's deployment environment, with OEM and on-premises support, dedicated support and SLAs, and a stated policy that customer data remains in the customer's environment and is not used to train models.
Run supported models on phones, laptops, or edge boxes using local formats such as GGUF, MLX, or ONNX when an application needs local inference rather than a hosted API.
Use the API or hosted access options to add model inference to software without operating the model infrastructure directly. The exact endpoint, billing, and model availability depend on the selected platform.
Evaluate text, vision-language, audio, or decision models for device-aware products such as vehicle, robotics, wearable, or other edge-oriented systems, selecting the release that matches the required modality and hardware.
Download and fine-tune an open weight model for a domain workflow while keeping the resulting fine-tuned model private, subject to the LFM Open License and its commercial-use conditions.
Work with Liquid AI on a model tailored to a specific environment, with options described for OEM and on-premises deployment, dedicated support and SLAs, and customer-controlled data environments.
Liquid models can be accessed through the Liquid API, with the API page stating that the endpoint is OpenAI-compatible. Models are also available through Liquid Console, OpenRouter, and Vercel AI Gateway, subject to each platform's billing and account setup.
Liquid's open weight models may be used commercially at no cost while the company's annual revenue is under $10 million. Once annual revenue exceeds that threshold, the organization must contact Liquid AI for a paid commercial license. Research, education, and nonprofit use is stated to remain free without a revenue limit.
The LFM Open License permits users to modify and fine-tune models without a copyleft requirement, so fine-tuned models can remain private. Original copyright and license notices must be retained when distributing the model or derivatives.
Liquid describes local deployment options for laptops, phones, and edge boxes, including GGUF, MLX, and ONNX builds. It also lists GPU-serving options such as vLLM, SGLang, Transformers, Baseten, Modal, and OpenRouter; availability depends on the specific model and deployment path.
The pricing page lists API access, open-weight model licensing, and custom deployment as separate options. Custom engagements can include model optimization for a deployment environment, on-premises or OEM support, dedicated support and SLAs, and keeping customer data within the customer's environment.
AakarDev AI helps teams manage AI provider access, project-level setups, logs, and analytics from one dashboard. It supports BYOK workflows and lists providers including OpenAI, Google Gemini, Anthropic, Groq, Mistral AI, and Perplexity AI.
CreateOS Sandbox is an isolated compute environment for running code and agent workloads inside Firecracker micro-VMs. It is designed for workflows that need machine-level isolation, private networking between sandboxes, and programmatic control through SDK, CLI, or MCP.
GitBot is a local browser-based workspace for creating reusable coding bots powered by Claude Code, Codex, or OpenCode. It helps developers run those agents across repository folders with persistent threads and configurable permissions.
Trigger.dev chat agent is a durable AI chat backend for developers building stateful conversations that can survive refreshes, crashes, and long-running turns. It connects with the AI SDK `useChat` flow and runs on managed infrastructure with no timeout on a turn.
ByteAsk is a terminal-first AI coding agent for C and C++ that edits repositories and verifies changes with the real compiler, debugger, sanitizers, and tests before showing a diff. It offers a free tier plus paid plans, with editor connectors and zero-retention handling described in the source.
Epherem is a static security scanner for web apps built with AI coding tools such as Lovable, Bolt, v0, Replit, Cursor, and Claude Code. It checks source files and dependencies for common security mistakes, then provides categorized findings and suggested fixes without running the app.