Is unsloth/LFM2.5-8B-A1B-GGUF good for agentic (tool-using) tasks?

In our benchmarks, unsloth/LFM2.5-8B-A1B-GGUF scores 67.3/100 for agentic workflows. It runs at about 144.9 tok/s, so if you want more speed, unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-Q4_K_XL is faster (~175.3 tok/s) and still scores well for agentic workflows (83.6/100).

How fast is unsloth/LFM2.5-8B-A1B-GGUF for local inference?

Across 4 community benchmark runs, unsloth/LFM2.5-8B-A1B-GGUF reaches up to 241.2 tok/s and averages 144.9 tok/s, with the fastest results on NVIDIA GeForce RTX 4070.

How much memory does unsloth/LFM2.5-8B-A1B-GGUF need?

The leanest observed configuration used about 3.9 GB of memory.

Which tools have been used to run unsloth/LFM2.5-8B-A1B-GGUF?

Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.

Model benchmarks

unsloth/LFM2.5-8B-A1B-GGUF local LLM performance

As of September 2026, unsloth/LFM2.5-8B-A1B-GGUF runs at up to 241.2 tok/s for local inference (best of 4 community benchmark runs across 2 GPUs).

llama.cpp

ShareReddit X

Model size

Peak speed

241.2 tok/s

Average speed

144.9 tok/s

Avg PP

699.8 tok/s

Min memory

3.9 GB

Max context

32.768 tokens

Avg output / run

17.678 tokens

Avg runtime / run

3m 45s

Avg quality

47.1

Benchmark runs

GPUs tested

Quality by task

Average LLM-judged quality (0–100) with the run-to-run spread shown as a P5–P95 band, overall and for each benchmark task, across all 4 runs. The low and high columns show how much the judge’s score varies between runs, and need at least two runs to display.

Task	P5 (low)	Avg	P95 (high)
Overall	41.7	47.1	52.2
Agent Workflow	50.5	67.3	81.9
Code Generation	10.2	14.8	19.5
Role Play & Narrative	54.8	64.7	75.4
Research & Analysis	36.0	41.6	47.1

Performance by hardware and tool

Every hardware/tool/quantization combination unsloth/LFM2.5-8B-A1B-GGUF has been benchmarked on, ranked by peak token generation speed. Last updated September 2026.

Hardware	Tool	Quant	Best tok/s	Avg tok/s	Memory	Context	Quality	Runs
NVIDIA GeForce RTX 4070	llama.cpp	—	241.2 tok/s	240.5 tok/s	3.9 GB	32.768 tokens	48.7	2
Intel(R) Arc(TM) 140V GPU (16GB)	llama.cpp	—	50.6 tok/s	49.3 tok/s	3.9 GB	32.768 tokens	45.5	2

Benchmark runs

All 4 unsloth/LFM2.5-8B-A1B-GGUF runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.

Frequently asked questions

Is unsloth/LFM2.5-8B-A1B-GGUF good for agentic (tool-using) tasks?: In our benchmarks, unsloth/LFM2.5-8B-A1B-GGUF scores 67.3/100 for agentic workflows. It runs at about 144.9 tok/s, so if you want more speed, unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-Q4_K_XL is faster (~175.3 tok/s) and still scores well for agentic workflows (83.6/100).
How fast is unsloth/LFM2.5-8B-A1B-GGUF for local inference?: Across 4 community benchmark runs, unsloth/LFM2.5-8B-A1B-GGUF reaches up to 241.2 tok/s and averages 144.9 tok/s, with the fastest results on NVIDIA GeForce RTX 4070.
How much memory does unsloth/LFM2.5-8B-A1B-GGUF need?: The leanest observed configuration used about 3.9 GB of memory.
Which tools have been used to run unsloth/LFM2.5-8B-A1B-GGUF?: Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.