How fast is prism-ml/Ternary-Bonsai-2-4B-gguf:PTQ1_0 for local inference?

Across 2 community benchmark runs, prism-ml/Ternary-Bonsai-2-4B-gguf:PTQ1_0 reaches up to 252.6 tok/s and averages 251.5 tok/s, with the fastest results on NVIDIA GeForce RTX 4070.

How much memory does prism-ml/Ternary-Bonsai-2-4B-gguf:PTQ1_0 need?

The leanest observed configuration used about 7.5 GB of memory (quantizations tested: Q1_0).

Which tools have been used to run prism-ml/Ternary-Bonsai-2-4B-gguf:PTQ1_0?

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

Model benchmarks

prism-ml/Ternary-Bonsai-2-4B-gguf:PTQ1_0 local LLM performance

As of October 2026, prism-ml/Ternary-Bonsai-2-4B-gguf:PTQ1_0 runs at up to 252.6 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).

llama.cppQ1_0

ShareReddit X

Model size

Peak speed

252.6 tok/s

Average speed

251.5 tok/s

Avg PP

4167.3 tok/s

Min memory

7.5 GB

Max context

262,144 tokens

Avg output / run

17,813 tokens

Avg runtime / run

1m 14s

Avg quality

53.3

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 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.

Task	P5 (low)	Avg	P95 (high)
Overall	52.4	53.3	54.2
Agent Workflow	63.6	70.4	77.2
Code Generation	10.1	19.8	29.5
Role Play & Narrative	61.1	65.9	70.7
Research & Analysis	53.0	57.1	61.3

Performance by hardware and tool

Every hardware/tool/quantization combination prism-ml/Ternary-Bonsai-2-4B-gguf:PTQ1_0 has been benchmarked on, ranked by peak token generation speed. Last updated October 2026.

Hardware	Tool	Quant	Best tok/s	Avg tok/s	Memory	Context	Quality	Runs
NVIDIA GeForce RTX 4070	llama.cpp	Q1_0	252.6 tok/s	251.5 tok/s	7.5 GB	262,144 tokens	53.3	2

Benchmark runs

All 2 prism-ml/Ternary-Bonsai-2-4B-gguf:PTQ1_0 runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.

Open run page →

Frequently asked questions

How fast is prism-ml/Ternary-Bonsai-2-4B-gguf:PTQ1_0 for local inference?: Across 2 community benchmark runs, prism-ml/Ternary-Bonsai-2-4B-gguf:PTQ1_0 reaches up to 252.6 tok/s and averages 251.5 tok/s, with the fastest results on NVIDIA GeForce RTX 4070.
How much memory does prism-ml/Ternary-Bonsai-2-4B-gguf:PTQ1_0 need?: The leanest observed configuration used about 7.5 GB of memory (quantizations tested: Q1_0).
Which tools have been used to run prism-ml/Ternary-Bonsai-2-4B-gguf:PTQ1_0?: Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.