Model benchmarks
Ternary-Bonsai-2-27B-PTQ1_0 local LLM performance
As of November 2026, Ternary-Bonsai-2-27B-PTQ1_0 runs at up to 57.5 tok/s for local inference (best of 7 community benchmark runs across 3 GPUs).
Model size
27B
Peak speed
57.5 tok/s
Average speed
41.7 tok/s
Avg PP
457.3 tok/s
Min memory
11.4 GB
Max context
128,000 tokens
Avg output / run
69,609 tokens
Avg runtime / run
33m 50s
Avg quality
81.4
Benchmark runs
7
GPUs tested
3
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 7 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 | 75.5 | 81.4 | 88.2 |
| Agent Workflow | 71.4 | 84.9 | 90.8 |
| Code Generation | 45.2 | 64.8 | 84.7 |
| Role Play & Narrative | 88.2 | 91.3 | 92.8 |
| Research & Analysis | 81.1 | 84.8 | 88.2 |
Performance by hardware and tool
Every hardware/tool/quantization combination Ternary-Bonsai-2-27B-PTQ1_0 has been benchmarked on, ranked by peak token generation speed. Last updated November 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| NVIDIA GeForce RTX 4070 | OpenAI-compatible | Q1_0 | 57.5 tok/s | 56.2 tok/s | 13.8 GB | 65,536 tokens | 82.4 | 3 |
| NVIDIA GeForce RTX 4070 | llama.cpp | Q1_0 | 56.3 tok/s | 42.5 tok/s | 11.4 GB | 65,536 tokens | 80.4 | 2 |
| Tesla T4 | llama.cpp | Q1_0 | 30.5 tok/s | 30.5 tok/s | 12.1 GB | 32,768 tokens | 87.3 | 1 |
| Intel Graphics | llama.cpp | Q1_0 | 7.6 tok/s | 7.6 tok/s | n/a | 128,000 tokens | 74.7 | 1 |
Benchmark runs
All 7 Ternary-Bonsai-2-27B-PTQ1_0 runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is Ternary-Bonsai-2-27B-PTQ1_0 good for coding?
- In our benchmarks, Ternary-Bonsai-2-27B-PTQ1_0 scores 64.8/100 for coding. It runs at about 41.7 tok/s, so if you want more speed, Inkling-2-Mini-oQ4e-mtp is faster (~57.7 tok/s) and still scores well for coding (83.9/100).
- Is Ternary-Bonsai-2-27B-PTQ1_0 good for agentic (tool-using) tasks?
- In our benchmarks, Ternary-Bonsai-2-27B-PTQ1_0 scores 84.9/100 for agentic workflows. It runs at about 41.7 tok/s, so if you want more speed, Nex-N2.5-mini-APEX-Mini is faster (~88.2 tok/s) and still scores well for agentic workflows (91.3/100).
- How fast is Ternary-Bonsai-2-27B-PTQ1_0 for local inference?
- Across 7 community benchmark runs, Ternary-Bonsai-2-27B-PTQ1_0 reaches up to 57.5 tok/s and averages 41.7 tok/s, with the fastest results on NVIDIA GeForce RTX 4070.
- How much memory does Ternary-Bonsai-2-27B-PTQ1_0 need?
- The leanest observed configuration used about 11.4 GB of memory (quantizations tested: Q1_0).
- Which tools have been used to run Ternary-Bonsai-2-27B-PTQ1_0?
- Benchmarks were submitted using OpenAI-compatible, llama.cpp. Results are community-contributed and updated as new runs arrive.