# Ternary Bonsai 2 27B Benchmark Scores & Performance

> Ternary Bonsai 2 27B by Prism ML scores 50.78/100 overall, ranking #113 out of 490 AI models.

Canonical page: https://benchlm.ai/models/ternary-bonsai-2-27b

Last updated: September 18, 2026

## Model Details

| Property | Value |
|----------|-------|
| Creator | Prism ML |
| Source Type | Open Weight |
| Reasoning Type | Reasoning |
| Context Window | 262K |
| Official model card | [PrismML Ternary Bonsai 2 27B model card](https://huggingface.co/prism-ml/Ternary-Bonsai-2-27B-gguf) |
| Overall Score | 50.78/100 |
| Overall Rank | #113 of 490 |

## Family & Coverage

- Family: Ternary Bonsai 2
- Variant: 27b
- Benchmarks covered: 22 of 446
- Related earlier model: [Ternary Bonsai 27B](/models/ternary-bonsai-27b)
- Coverage note: BenchLM currently has partial benchmark coverage for this model, so the overall score is conservative.

## Agentic Benchmarks

| Benchmark | Score |
|-----------|-------|
| [τ²-bench results](/benchmarks/tau2-bench) | 80.2% |
| [BFCL v3](/benchmarks/bfclv3) | 74.9% |
| [Terminal-Bench 2.1](/benchmarks/terminalbench21) | 52.8% |

## Coding Benchmarks

| Benchmark | Score |
|-----------|-------|
| [LiveCodeBench v6](/benchmarks/livecodebench-v6) | 90.1% |
| [BigCodeBench](/benchmarks/bigcodebench) | 58.1% |
| [SWE-bench Verified](/benchmarks/swe-bench-verified) | 60.8% |
| [Terminal-Bench 2.1](/benchmarks/terminalbench21) | 52.8% |

## Multimodal & Grounded Benchmarks

| Benchmark | Score |
|-----------|-------|
| [CharXiv (overall)](/benchmarks/charxivoverall) | 80.0% |
| [A-OKVQA](/benchmarks/aokvqa) | 86.8% |
| [OmniDocBench 1.6](/benchmarks/omnidocbench16) | 89.1% |
| [RealWorldQA](/benchmarks/realworldqa) | 80.1% |
| [OCRBench V2](/benchmarks/ocrbenchv2) | 56.9% |

## Reasoning Benchmarks

| Benchmark | Score |
|-----------|-------|
| [AA-LCR](/benchmarks/lcr) | 77.0% |

## Knowledge Benchmarks

| Benchmark | Score |
|-----------|-------|
| [MMLU-Redux](/benchmarks/mmluredux) | 89.1% |
| [GPQA](/benchmarks/gpqa) | 85.8% |
| [GPQA-D](/benchmarks/gpqa-diamond) | 85.8% |

## Instruction Following Benchmarks

| Benchmark | Score |
|-----------|-------|
| [IFEval](/benchmarks/ifeval) | 91.3% |
| [IFBench](/benchmarks/ifbench) | 74% |

## Mathematics Benchmarks

| Benchmark | Score |
|-----------|-------|
| [GSM8K](/benchmarks/gsm8k) | 96.7% |
| [MATH-500](/benchmarks/math-500) | 98.8% |
| [AIME 2025](/benchmarks/aime2025) | 95% |
| [AIME26](/benchmarks/aime2026) | 95.8% |

## Other Prism ML Models

- [1-bit Bonsai 27B](/models/bonsai-27b) - Score: not computed
- [Ternary Bonsai 27B](/models/ternary-bonsai-27b) - Score: not computed
- [Ternary Bonsai 1.7B](/models/ternary-bonsai-1-7b) - Score: not computed
- [Ternary Bonsai 4B](/models/ternary-bonsai-4b) - Score: not computed
- [Ternary Bonsai 8B](/models/ternary-bonsai-8b) - Score: not computed
- [1-bit Bonsai 1.7B](/models/bonsai-1-7b) - Score: not computed
- [1-bit Bonsai 4B](/models/bonsai-4b) - Score: not computed
- [1-bit Bonsai 8B](/models/bonsai-8b) - Score: not computed