Model benchmarks

Ternary-Bonsai-2-27B-PQ2_0-MTP-Q8_0 local LLM performance

As of September 2026, Ternary-Bonsai-2-27B-PQ2_0-MTP-Q8_0 runs at up to 61.3 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).

llama.cppQ2_0
ShareRedditX

Model size

27B

Peak speed

61.3 tok/s

Average speed

60.6 tok/s

Avg PP

548.9 tok/s

Min memory

26.4 GB

Max context

262.144 tokens

Avg output / run

79.828 tokens

Avg runtime / run

23m 6s

Avg quality

83.7

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)
Overall82.183.785.3
Agent Workflow80.581.482.3
Code Generation76.878.480.0
Role Play & Narrative86.891.195.3
Research & Analysis83.483.984.3

Performance by hardware and tool

Every hardware/tool/quantization combination Ternary-Bonsai-2-27B-PQ2_0-MTP-Q8_0 has been benchmarked on, ranked by peak token generation speed. Last updated September 2026.

HardwareToolQuantBest tok/sAvg tok/sMemoryContextQualityRuns
AMD Radeon Pro W7900llama.cppQ2_061.3 tok/s60.6 tok/s26.4 GB262.144 tokens83.72

Benchmark runs

All 2 Ternary-Bonsai-2-27B-PQ2_0-MTP-Q8_0 runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.

Frequently asked questions

Is Ternary-Bonsai-2-27B-PQ2_0-MTP-Q8_0 good for coding?
In our benchmarks, Ternary-Bonsai-2-27B-PQ2_0-MTP-Q8_0 scores 78.4/100 for coding. It runs at about 60.6 tok/s, so if you want more speed, unsloth/Qwen3.8-27B-GGUF:IQ3_S is faster (~104.3 tok/s) and still scores well for coding (83.0/100).
Is Ternary-Bonsai-2-27B-PQ2_0-MTP-Q8_0 good for agentic (tool-using) tasks?
In our benchmarks, Ternary-Bonsai-2-27B-PQ2_0-MTP-Q8_0 scores 81.4/100 for agentic workflows. It runs at about 60.6 tok/s, so if you want more speed, granite-4.2-8b-Q4_K_L is faster (~71.6 tok/s) and still scores well for agentic workflows (92.9/100).
How fast is Ternary-Bonsai-2-27B-PQ2_0-MTP-Q8_0 for local inference?
Across 2 community benchmark runs, Ternary-Bonsai-2-27B-PQ2_0-MTP-Q8_0 reaches up to 61.3 tok/s and averages 60.6 tok/s, with the fastest results on AMD Radeon Pro W7900.
How much memory does Ternary-Bonsai-2-27B-PQ2_0-MTP-Q8_0 need?
The leanest observed configuration used about 26.4 GB of memory (quantizations tested: Q2_0).
Which tools have been used to run Ternary-Bonsai-2-27B-PQ2_0-MTP-Q8_0?
Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.