LiquidAI/ lfm2.5-2.6b:f16

318 5 days ago

Fast 2.6B local AI agent with thinking that rivals models 4x larger on tool use and multi-step agentic tasks.

tools thinking 
ollama run LiquidAI/lfm2.5-2.6b:f16

Details

5 days ago

b1518c247ce5 · 5.4GB

model
lfm2
·
2.7B
·
F16
license
LFM Open License v1.0 TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION 1. Definitions. "
params
{ "repeat_penalty": 1.1, "temperature": 0.1, "top_k": 50 }

Readme

Liquid AI

LFM2.5-2.6B

LFM2.5-2.6B is part of LFM2.5, a family of hybrid models designed for on-device deployment. It builds on the LFM2 architecture with a 128K context window and agentic post-training.

  • Best-in-class agent: Competitive with models 4x larger on tool use, instruction following, and multi-step agentic tasks.
  • Agentic reinforcement learning: Trained inside the most popular agentic harnesses to improve compatibility.
  • Efficient inference: 220 tok/s on an Apple M5 Max and 113 tok/s on an AMD Ryzen CPU, in under 2.5 GB of memory.

Find more information about LFM2.5-2.6B in our blog post.

🏃 Run with Ollama

ollama run LiquidAI/lfm2.5-2.6b

Available tags: bf16, f16, q4_0, q4_k_m (latest), q5_k_m, q6_k, q8_0.

Each tag is built from the LFM2.5-2.6B-GGUF GGUF files with Ollama’s native LFM2 renderer and parser, so tool calling works out of the box, and carries the checkpoint’s recommended sampling parameters. The full model card (chat template, tool-use format, inference and fine-tuning guides) is on Hugging Face.

🗒️ Model Details

LFM2.5-2.6B is a general-purpose text-only model with the following features:

  • Total parameters: 2.69B
  • Number of layers: 30 (22 double-gated short convolution blocks + 8 GQA)
  • Training budget: 34 trillion tokens
  • Vocabulary size: 128,000
  • Context length: 131,072 tokens
  • Languages: English, Arabic, Chinese, French, German, Italian, Japanese, Korean, Portuguese, Spanish, Vietnamese, Thai, Indonesian, Hindi, Russian, Polish
  • Generation parameters:
    • temperature: 0.1
    • top_k: 50
    • repetition_penalty: 1.1

We recommend using it for agentic workloads, tool use, data extraction, RAG, and long-context workflows. It is not recommended for agentic coding and knowledge-heavy tasks.

Training

LFM2.5-2.6B is pre-trained on ~34T tokens, with a mid-training phase that extends the context window to 128K. Post-training then turns the base model into an agent in four stages: supervised fine-tuning (two rounds), per-domain teacher specialization, multi-domain on-policy distillation, and agentic reinforcement learning.

In particular, agentic reinforcement learning allows us to directly train the model inside popular agentic harnesses. It exposes the model to their tools, system prompts, and interaction patterns, helping it work reliably across agent environments.

📊 Performance

Benchmarks

We compared LFM2.5-2.6B with relevant sub-10B models on a diverse suite of benchmarks.

Benchmark LFM2.5-2.6B (2.6B) gemma-4-E2B-it (5.1B) gemma-4-E4B-it (8B) Qwen3.5-4B (4.7B) Qwen3.5-9B (9.7B)
AA-Omni-Public Index -29.50 -74.47 -49.03 -54.30 -50.43
AA-Omni-Public Acc 8.13 6.37 8.33 17.63 21.30
AA-Omni-Public Non-hallu 59.04 13.67 37.42 12.66 8.84
AIME25 51.87 26.33 34.27 49.33 56.07
LiveCodeBenchv6 59.41 54.92 63.77 60.85 69.86
IFBench 59.17 34.08 39.24 48.40 56.47
Multi-IF 80.07 69.44 77.35 55.67 62.55
IFStruct 85.49 64.85 76.65 36.25 78.50
BFCLv4 56.88 36.98 46.39 50.56 60.13
ToolSandbox 77.83 52.40 65.00 75.55 76.44
τ³-Bench Banking 5.67 3.35 4.12 5.45 5.15
Claw-Eval average (EN) 62.85 53.14 58.02 62.28 66.53
PinchBench 68.22 44.24 55.09 71.26 71.45
BrowseComp+ (OpenClaw) 26.89 8.31 15.90 24.46 27.23

CPU Inference

Due to its efficient LFM2 architecture, LFM2.5-2.6B is the fastest model we tested, with decode speeds of 220 tokens/s on an M5 Max and 113 tokens/s on a Ryzen AI Max+ 395. At 30 tokens/s, it allows you to run capable agents even on a phone.

GPU Inference

LFM2.5-2.6B is the fastest model in its size class, reaching almost 15K output tokens per second at high concurrency, roughly 1.3B tokens per day on a single H100.

📬 Contact