liquid
LiquidAI: LFM2.5-24B-A2B
liquid/lfm-2.5-24b-a2b
Our take
AI-assisted analysis based on the sources below. Suitability is an editorial assessment; test it on your own workload.
Liquid AI's LFM2.5-24B-A2B is a text-only mixture-of-experts reasoning model with 24B total and 2.45B active parameters, post-trained for agentic workloads. Per its model card and the OpenRouter catalog, it always produces a reasoning pass before answering, supports function calling and structured outputs, and offers a context window of roughly 128K-131K tokens across 16 listed languages. Vendor-reported results emphasize tool use and on-device inference speed; no independent evaluations were included in this source pack.[1][2]
Strengths
- Sparse MoE design activates only 2.45B of 24B parameters per token, which the vendor says keeps each reasoning token inexpensive; the card reports high throughput on both Apple Silicon and data-center GPUs.[2]
- Built for agents: reasoning is mandatory before every answer, the card documents a four-step function-calling flow, and post-training reportedly included agentic reinforcement learning inside popular agent harnesses.[1][2]
- Broad deployment surface, with vendor-published FP8, GGUF, ONNX, and MLX checkpoints plus support for Transformers, vLLM, SGLang, llama.cpp, MLX, and LM Studio; OpenRouter lists tools, tool choice, and structured outputs among supported parameters.[1][2]
Limitations
- The vendor itself advises against relying on it alone for knowledge-heavy tasks and recommends pairing it with retrieval or web search; the knowledge cutoff is not stated in the evidence.[1][2]
- Capability, comparison, and speed claims are vendor-reported; the supplied benchmark table is truncated and no independent evaluations appear in this source pack, so workload-specific validation is still needed.[1][2]
- Reported context length differs between sources (131,072 tokens in the model card versus 128,000 in the OpenRouter catalog), and the model ships under a custom lfm1.0 license whose terms are not detailed in the evidence.[1][2]
Where to use it
- On-device or self-hosted agents that must reason and call tools where privacy and latency matter; a suggested fit based on vendor positioning, not a proven result.[2]
- Function-calling pipelines served through OpenAI-compatible endpoints, since tools, tool choice, response format, and structured outputs appear among the catalog's supported parameters.[1][2]
- Multilingual or retrieval-augmented assistants across the 16 card-listed languages, leaning on tool use rather than stored knowledge.[2]
Sources
- [1] OpenRouter catalog · LiquidAI: LFM2.5-24B-A2BSources checked: Nov 13, 2026
- [2] Model documentation · LiquidAI: LFM2.5-24B-A2BSources checked: Nov 13, 2026
About this model
Description below is the original text supplied by OpenRouter, shown in the source language.
LFM2.5-24B-A2B is a mixture-of-experts reasoning model from Liquid AI with 24B total and 2.45B active parameters. It always reasons before answering and is built for agents that call tools. Liquid...
Specifications
- Model ID
- liquid/lfm-2.5-24b-a2b
- Canonical ID
- liquid/lfm-2.5-24b-a2b-20261012
- Creator
- liquid
- Added to OpenRouter
- 13.10.2026
- Context length
- 128,000
- Modality
- text->text
- Input modalities
- text
- Output modalities
- text
- Token pricing
- USD per 1M tokens: in $0.050 / out $0.700
- Knowledge cutoff
- —
- Expiration date
- —
Tasks and capabilities
A grade appears only where the source supplies a matching measurement for that task, the model is active and the cohort holds at least 10 comparable models. The capability column lists facts the source declares — supported formats, reasoning and tool parameters — which are not measured quality.
| Task | Grade | Source-declared capability |
|---|---|---|
| General quality | No measurement | reasoning: mandatory · parameters: include_reasoning, reasoning |
| Coding | No measurement | — |
| Agents | No measurement | parameters: structured_outputs, tool_choice, tools |
| Writing | No measurement | output: text |
| Reasoning | No measurement | reasoning: mandatory · parameters: include_reasoning, reasoning |
| Document analysis | No measurement | — |
| Vision | No measurement | — |
| Image generation | No measurement | — |
| Audio | No measurement | — |
| Video | No measurement | — |
| Embeddings | No measurement | — |
Evaluations
No evaluations are supplied for this model.
How grades work
Each letter grade compares a model only against other models scored by the same source and metric in the current snapshot. S means the top 10%, A the next 20%, B the next 30%, C the rest. Models without a published measurement are shown as not rated — a supported format alone never implies measured quality.
- S — S — top 10% of the cohort
- A — A — next 20% of the cohort
- B — B — next 30% of the cohort
- C — C — below the B band
Providers
Optional live provider statistics from OpenRouter. Missing values are unknown, not zero.
| Provider | Context length | Price in /1M | Price out /1M | Latency p50 (30m, s) | Throughput p50 (30m, tok/s) | Uptime (30m) |
|---|---|---|---|---|---|---|
| Liquidfp8 | 128,000 | $0.050 | $0.700 | — | — | — |
Provider data retrieved: 13.11.2026
Related articles
No published articles identify this model yet.