Liquid LFM2.5 24B
Liquid AI's mixture-of-experts reasoning model with 24B total and 2.45B active parameters. It always reasons before answering and is built for agents that call tools.
From Liquid, LFM2.5 24B is a cloud text model. It takes a text prompt and replies with generated text. Its context window handles up to 128K input tokens and up to 8K output tokens. Generation can be tuned with system instructions and temperature. It runs through OpenRouter using your own API key, from $0.05 per million input tokens. It also runs on your own hardware through Ollama, with no API key and no per-use fees.
- Released
- Oct 2026
- Pricing
- $0.05 / 1M in · $0.70 / 1M out
- Context window
- 128K in · 8K out
- Controls
- System prompt, Temperature
Samples
Generated with LFM2.5 24B via OpenRouter. The same three prompts run against every text model in the catalog, shown verbatim, so the only thing that changes between two models’ answers is the model.
Explain what an API is to someone who has never written code. Do not use a restaurant or waiter analogy. Plain prose only — no headings, bullet points, or markdown formatting. Maximum 50 words.
An API is a collection of instructions that lets one piece of software ask another for specific help. It describes the format for requests and expected answers, so programs can work together without needing to understand each other's inner workings.
A two-person startup is deciding whether to build their own user authentication or use a hosted provider. Give them a clear recommendation, then the single strongest argument against your own recommendation. Plain prose only — no headings, bullet points, or markdown formatting. Maximum 150 words.
For a two‑person startup, the safest and fastest path is to adopt a hosted authentication service such as Auth0, Firebase Auth, or AWS Cognito rather than building your own system. This lets you ship quickly, offload security and compliance burdens, and concentrate engineering effort on your product. The strongest argument against this choice is that a custom solution gives you complete control over user data, allows deep integration of unique business logic, and eliminates reliance on a third‑party vendor’s uptime, pricing changes, or policy shifts, which can become critical as you scale.
Write a briefing for a non-technical executive explaining why their company's spending on AI is hard to predict, and what they can do about it. Cover at least three distinct causes. Plain prose only — no headings, bullet points, or markdown formatting. Write exactly 205 words.
Your company’s AI budget is difficult to forecast because the technology landscape changes faster than most traditional planning cycles. Breakthroughs in algorithms, new model architectures, and open‑source tools can shift the competitive frontier overnight, making it hard to know which capabilities will become essential within a fiscal year. Second, the business impact of AI projects is often ambiguous; the expected return on investment depends on factors such as user adoption, process automation, and regulatory approval that are uncertain until later stages. Third, the quality and accessibility of data that feeds AI systems vary widely across departments, and building pipelines requires upfront investment that is rarely captured in standard cost models. These three forces combine to create a moving target for finance teams. To regain control you can introduce a staged funding model that ties spend to validated pilots rather than speculative promises. Define measurable success criteria before each project begins, so progress can be tracked and budgets reallocated if results fall short. Invest in data governance and infrastructure, treating data readiness as a deliverable rather than an afterthought. Finally, assemble an oversight board that includes technical, legal, and commercial leaders to review assumptions regularly and adjust the roadmap as the technology and market conditions evolve.
Liquid
Liquid develops Liquid Foundation Models (LFM) — a new architecture optimized for efficient long-context reasoning.
www.liquid.ai ↗Liquid's structured decision model, served as a System One endpoint for choice, score and yes/no answers.
Liquid AI's largest decision model, served as a System One endpoint: it reads text and images and answers typed questions with a choice, a score or a yes/no probability.
Liquid's 24B foundation model with efficient long-context reasoning.
Liquid's efficient 8B Liquid Foundation Model tuned for long-context reasoning on-device.
Liquid AI's 600M decision model on LFM2.5-Encoder-350M, running fully offline: it answers typed questions about a text or JSON state with a choice, a score or a yes/no probability.