LFM 2.5 24B A2B
Neural Network
LFM 2.5 24B A2B is a reasoning MoE model by Liquid AI for agents with tool calling, available via API in BotHub.
Specifications
- Input
- text, documents
- Output
- text
- Context size
- 128 000 tokens
- Max answer length
- 8 192 tokens
- Reasoning effort
- low, medium, high
- Capabilities
- reasoning, function calling
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works LFM 2.5 24B A2B?
LFM 2.5 24B A2B is a reasoning model from Liquid AI with a mixture-of-experts architecture: out of 24 billion parameters, about 2.45 billion are active at each step, combining the knowledge of a large network with efficient computation. It always performs a reasoning step before responding. Its primary use case, as stated by the developer, is for agents that call external tools. In BotHub, the model is accessible from Russia without a VPN or foreign card: payment is made with Russian cards in rubles, charged per token, and unused balances do not expire. With over 250 models available in the same window, you can easily compare LFM 2.5 24B A2B with others and switch without rewriting your integration, thanks to the unified OpenAI-compatible API. Chats are encrypted, and for companies, we offer contracts, invoices, and an admin panel with limits. Use this model where you need an agent with function calling: data collection via external services, multi-step automation scenarios, task analysis with step-by-step logic, assistant prototypes with tool access, and backend for multimodal products.Frequently asked questions about LFM 2.5 24B A2B
You can use generated results for commercial purposes. You own all rights to the content you create. The only restriction: make sure your prompt does not include copyrighted third-party material. You are responsible for respecting the rights to any input data.
This is a Liquid text model with a 128,000 token context. It accepts text and documents, can reason before responding, and call external functions. It is suitable for parsing long files, working with code, extracting structured data, and analytical tasks within your services.
The model generates up to 8,192 tokens per response, which is several pages of text. If you need a large document or a large code file, ask for a continuation or split the task into parts: the 128,000 token context allows you to keep the entire conversation intact.
Yes, the reasoning mode is confirmed by our data. The model first builds a chain of steps and then formulates a conclusion — this helps with math, logic, multi-step instructions, and code analysis. Note that reasoning also consumes the 8,192 token limit per response.
Yes, function calling and structured JSON output are supported. The model can be connected to search, databases, calculators, and your own services, and you can receive responses strictly according to a specified schema. In BotHub, this works via a unified OpenAI-compatible API, without rewriting the integration when switching models.
In addition to text, the model accepts documents: upload a file and ask for a summary, version comparison, or extraction of specific fields. Images, audio, and video are not listed as input formats in our data, so for such tasks, it is more convenient to switch to a specialized model in the same window.
Our data does not specify the language of prompts and responses, so we cannot make specific promises — it is easier to test it with a couple of queries on your task. BotHub itself is accessible from Russia without a VPN or foreign card, and there are over 250 models in the same window for comparison.