# Ternary Bonsai 2 27B

Prism ML · Open Weight · rank 114 · bench-align-v5 · Self-hosted

> Every figure below is reproduced as its upstream source published it: nothing is modelled, estimated, interpolated or converted. `Unavailable` means no source published the value — it is never a zero. Each value carries its evidence state and the date it was observed.

Page: https://modelchoice.dev/models/ternary-bonsai-2-27b  
JSON: https://modelchoice.dev/api/model/ternary-bonsai-2-27b

## Facts

| Field | Value | Evidence |
| --- | --- | --- |
| Canonical id | `ternary-bonsai-2-27b` | — |
| Overall score | 46 | Observed 2026-11-21 · source benchlm:models |
| Verified score (BenchLM): a separate BenchLM figure, not the overall score | 66 | Observed 2026-11-21 · source benchlm:models |
| Context window | 262K tokens | Observed 2026-11-21 · source benchlm:models |
| Modalities (models.dev) | Input: text, image · Output: text | Observed 2026-11-21 · source openrouter:pricing |
| Release date | 2026-09-17 | Observed 2026-11-21 · source benchlm:models |
| Access type | Open Weight | — |
| Blended $/1M (75% input / 25% output) | $0.18 | Derived from the input and output rates below |

## Capability evidence

Seven axes from the ranking source. An axis the source did not score is unavailable, not zero. Beside each score is BenchLM's separate verified score for that axis; the two are published side by side and are not interchangeable.

| Axis | Score | Evidence | Verified (BenchLM) | Verified evidence |
| --- | --- | --- | --- | --- |
| Agentic | Unavailable | Unavailable · source benchlm:models | Unavailable | Unavailable · source benchlm:models |
| Coding | 43.6 | Observed 2026-11-21 · source benchlm:models | 43.5 | Observed 2026-11-21 · source benchlm:models |
| Knowledge | 75.3 | Observed 2026-11-21 · source benchlm:models | 75.3 | Observed 2026-11-21 · source benchlm:models |
| Reasoning | 77 | Observed 2026-11-21 · source benchlm:models | 77 | Observed 2026-11-21 · source benchlm:models |
| Multimodal & Grounded | Unavailable | Unavailable · source benchlm:models | Unavailable | Unavailable · source benchlm:models |
| Instruction Following | 74.8 | Observed 2026-11-21 · source benchlm:models | 74.8 | Observed 2026-11-21 · source benchlm:models |
| Math | 82.9 | Observed 2026-11-21 · source benchlm:models | 82.9 | Observed 2026-11-21 · source benchlm:models |

## Runtime service evidence

Measured values with the date they were observed. Nothing is inferred from a sibling model or a provider claim. Regional or per-endpoint measurements appear only when the API supplies them; none are modelled.

| Measurement | Value | Observed | Last good | Evidence |
| --- | --- | --- | --- | --- |
| Time to first token | Unavailable | Unavailable | Unavailable | Unavailable · source benchlm:speed |
| Throughput | 46 tok/s | 2026-11-18 | 2026-11-18 | Last good 2026-11-18 · source openrouter:endpoints |

## Endpoint and price matrix

Every published price component, including cache reads and writes.

| Component | USD | Evidence |
| --- | --- | --- |
| Input / 1M tokens | $0.075 | Observed 2026-11-21 · source openrouter:pricing |
| Output / 1M tokens | $0.50 | Observed 2026-11-21 · source openrouter:pricing |
| Cache read / 1M tokens | $0.0375 | Observed 2026-11-21 · source openrouter:pricing |
| Cache write / 1M tokens | Unavailable | Unavailable · source openrouter:pricing |
| Blended / 1M (75% input / 25% output) | $0.18 | Derived — from the input and output rates above; it has no source record of its own |
| Cost per successful task (LiveBench) | Unavailable | Unavailable · source livebench:table |

**Self-hosted listing.** Prism ML publishes Ternary Bonsai 2 27B under Apache 2.0 for self-hosting and does not publish a first-party hosted token rate for this exact checkpoint. We represent the open-weight row as self-host/free-per-token before infrastructure costs. The rates above are a hosted price matched from another provider, not a first-party list price.

## Workload-aware monthly cost example

10 conversations per day × 8 messages × 22 active days, 1200 input and 400 output tokens per message, no cache. Derived here from the published rates above by this site's own calculator — not a figure any source published.

| Field | Value |
| --- | --- |
| Modelled monthly cost | $0.51 |
| Modelled tokens | 2.82M |

## Benchmark record

22 matched benchmark rows with their published value, unit, and provenance.

### Knowledge

| Benchmark | Value | Tasks | Difficulty | Provenance |
| --- | --- | --- | --- | --- |
| GPQA (Graduate-Level Google-Proof Q&A) | 85.76 | 448 questions | Graduate level | [GPQA: A Graduate-Level Google-Proof Q&A Benchmark](https://arxiv.org/abs/2311.12022) |
| GPQA-D (GPQA Diamond) | 85.8 | Graduate-level science questions | Graduate level | [GPQA: A Graduate-Level Google-Proof Q&A Benchmark](https://arxiv.org/abs/2311.12022) |
| MMLU-Redux | 89.09 | Broad academic QA | Advanced general knowledge | [Qwen3.6 launch benchmarks](https://qwen.ai/blog?id=qwen3.6) |

### Coding

| Benchmark | Value | Tasks | Difficulty | Provenance |
| --- | --- | --- | --- | --- |
| Terminal-Bench 2.1 | 52.8 | Terminal-based software-agent tasks | Professional software engineering | [Terminal-Bench](https://www.tbench.ai/) |
| BigCodeBench | 58.1 | Code generation tasks | Software engineering | [DeepSeek-V4 Technical Report](https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro/blob/main/DeepSeek_V4.pdf) |
| SWE-bench Verified (Software Engineering Benchmark Verified) | 60.8 | 500 verified issues | Professional software engineering | [SWE-bench: Can Language Models Resolve Real-World GitHub Issues?](https://arxiv.org/abs/2310.06770) |
| LiveCodeBench v6 | 90.1 | Fresh programming problems | Competitive programming level | [LiveCodeBench official repository and release documentation](https://github.com/LiveCodeBench/LiveCodeBench) |

### Mathematics

| Benchmark | Value | Tasks | Difficulty | Provenance |
| --- | --- | --- | --- | --- |
| AIME 2025 (American Invitational Mathematics Examination 2025) | 95 | 15 problems | High school olympiad level | [American Invitational Mathematics Examination](https://www.maa.org/math-competitions/aime) |
| GSM8K (Grade School Math 8K) | 96.7 | Grade-school math word problems | Grade-school math | [Training Verifiers to Solve Math Word Problems](https://arxiv.org/abs/2110.14168) |
| MATH-500 (MATH-500 Problem Set) | 98.8 | 500 problems | High school to undergraduate | [Measuring Mathematical Problem Solving With the MATH Dataset](https://arxiv.org/abs/2103.03874) |
| AIME26 (AIME 2026) | 95.8 | Competition math problems | Olympiad-style mathematics | [Qwen3.6 launch benchmarks](https://qwen.ai/blog?id=qwen3.6) |

### Reasoning

| Benchmark | Value | Tasks | Difficulty | Provenance |
| --- | --- | --- | --- | --- |
| AA-LCR (Artificial Analysis Long Context Reasoning) | 77.0 | Long-context reasoning tasks | Long-context reasoning | [Artificial Analysis model benchmarks](https://artificialanalysis.ai/models/grok-4-3) |

### Instruction Following

| Benchmark | Value | Tasks | Difficulty | Provenance |
| --- | --- | --- | --- | --- |
| IFEval (Instruction-Following Eval) | 91.31 | 541 prompts across 25 instruction types | Instruction precision | [Instruction-Following Evaluation for Large Language Models](https://arxiv.org/abs/2311.07911) |
| IFBench (Instruction Following Benchmark) | 74 | — | — | [BenchLM](https://benchlm.ai/benchmarks/ifbench) |

### Agentic

| Benchmark | Value | Tasks | Difficulty | Provenance |
| --- | --- | --- | --- | --- |
| Terminal-Bench 2.1 | 52.8 | Terminal-based software-agent tasks | Professional software engineering | [Terminal-Bench](https://www.tbench.ai/) |
| τ²-bench results (τ²-Bench Tool-Agent-User Evaluation) | 80.22 | Airline, retail, and telecom customer-service task sets | Dual-control customer-service workflows | [τ²-Bench: Evaluating Conversational Agents in a Dual-Control Environment](https://arxiv.org/abs/2506.07982) |
| BFCL v3 (Berkeley Function Calling Leaderboard v3) | 74.9 | Function-calling tasks | Advanced tool use | [The Berkeley Function Calling Leaderboard (BFCL): From Tool Use to Agentic Evaluation of Large Language Models](https://gorilla.cs.berkeley.edu/leaderboard.html) |

### Multimodal & Grounded

| Benchmark | Value | Tasks | Difficulty | Provenance |
| --- | --- | --- | --- | --- |
| OCRBench V2 | 56.9 | Image OCR tasks | Native visual text understanding | [OCRBench v2: An Improved Benchmark for Evaluating Large Multimodal Models on Visual Text Localization and Reasoning](https://arxiv.org/abs/2501.00321) |
| RealWorldQA | 80.1 | Real-world visual question answering | General visual reasoning | [Qwen3.6 launch benchmarks](https://qwen.ai/blog?id=qwen3.6) |
| CharXiv (overall) (CharXiv Descriptive and Reasoning Combined) | 80.0 | Scientific chart description and reasoning | Scientific visualization reasoning | [CharXiv: Charting Gaps in Realistic Chart Understanding in Multimodal LLMs](https://charxiv.github.io/) |
| A-OKVQA (A Benchmark for Visual Question Answering using World Knowledge) | 86.8 | Knowledge-grounded visual question answering | Commonsense and world knowledge about images | [A-OKVQA: A Benchmark for Visual Question Answering using World Knowledge](https://arxiv.org/abs/2206.01718) |
| OmniDocBench 1.6 (OmniDocBench v1.6) | 89.1 | Document parsing and extraction | Grounded document reasoning | [OmniDocBench: Benchmarking Diverse PDF Document Parsing with Comprehensive Annotations](https://arxiv.org/abs/2412.07626) |

## Lifecycle and limitations log

Lifecycle events the source associates with this model.

No lifecycle event references this model. That is not evidence the model has no lifecycle plan — only that this source published none.

## What this profile does not claim

Values are reproduced exactly as their sources published them, in the units those sources declared; none are converted, interpolated, or averaged across providers. Any field marked unavailable was attempted and not returned. Last attempted fetch for this model's score: 2026-11-21 06:17 UTC.
