Model benchmarks

laguna-xs-2.1 local LLM performance

As of July 2026, laguna-xs-2.1 runs at up to 37.2 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).

LM Studio
ShareRedditX

Model size

Unknown

Peak speed

37.2 tok/s

Average speed

27.4 tok/s

Min memory

18.9 GB

Max context

65,536 tokens

Avg quality

62.0

Benchmark runs

2

GPUs tested

1

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 2 runs. The low and high columns show how much the judge’s score varies between runs, and need at least two runs to display.

TaskP5 (low)AvgP95 (high)
Overall55.662.068.5
Agent Workflow78.981.984.8
Code Generation2.929.055.1
Role Play & Narrative66.370.875.3
Research & Analysis65.166.567.9

Performance by hardware and tool

Every hardware/tool/quantization combination laguna-xs-2.1 has been benchmarked on, ranked by peak token generation speed. Last updated July 2026.

HardwareToolQuantBest tok/sAvg tok/sMemoryContextQualityRuns
NVIDIA GeForce RTX 4070 Ti SUPERLM Studio37.2 tok/s27.4 tok/s18.9 GB65,536 tokens62.02

Frequently asked questions

Is laguna-xs-2.1 good for coding?
In our benchmarks, laguna-xs-2.1 scores 29.0/100 for coding. It runs at about 27.4 tok/s, so if you want more speed, K2-Horizon-MoVA-36B-A4B-MLX-4bit is faster (~48.8 tok/s) and still scores well for coding (85.3/100).
Is laguna-xs-2.1 good for agentic (tool-using) tasks?
In our benchmarks, laguna-xs-2.1 scores 81.9/100 for agentic workflows. It runs at about 27.4 tok/s, so if you want more speed, muse-glimmer:latest is faster (~34.5 tok/s) and still scores well for agentic workflows (93.0/100).
How fast is laguna-xs-2.1 for local inference?
Across 2 community benchmark runs, laguna-xs-2.1 reaches up to 37.2 tok/s and averages 27.4 tok/s, with the fastest results on NVIDIA GeForce RTX 4070 Ti SUPER.
How much memory does laguna-xs-2.1 need?
The leanest observed configuration used about 18.9 GB of memory.
Which tools have been used to run laguna-xs-2.1?
Benchmarks were submitted using LM Studio. Results are community-contributed and updated as new runs arrive.