Prism ML
availableShows if the model has enough results for an index.Ternary Bonsai 2 27B
Ternary Bonsai 2 27B is a reasoning model from Prism ML in the Ternary Bonsai 2 family. 13 benchmarks count toward its score, in 7 categories.
IndexOverall score out of 100.52.7 ±5.0
CoverageShare of the index weight with results.95%
SpeedOutput tokens per second.10/s
Input / 1MUS dollars per 1M input tokens.$0.075
Output / 1MUS dollars per 1M output tokens.$0.5
ContextMaximum tokens in one request.262K
EloLMArena rating and rank.N/A
The index is a score out of 100. The ± range shows how much it can change.
CapabilitiesScore per category, out of 100.
Out of 100Results
13 counted| BenchmarkThe test name. | CategoryThe capability that the test measures. | ResultThe score from the publisher. | IndexThis result as a score out of 100. | RunThe settings of the run. | DateDate of the result. | Published byThe source of the result. |
|---|---|---|---|---|---|---|
| MATH-500 Problem Set | Math | 98.8% | 52.2 | — | — | Dan Hendrycks et al. |
| Grade School Math 8K | Math | 96.7% | — | — | — | DeepSeek-AI |
| AIME 2026 | Math | 95.8% | 54.8 | — | — | Qwen |
| American Invitational Mathematics Examination 2025 | Math | 95.0% | 51.4 | — | — | Mathematical Association of America |
| Instruction-Following Eval | Instruction | 91.3% | 48.0 | — | — | Jeffrey Zhou et al. |
| LiveCodeBench v6 | Coding | 90.1% | 59.0 | — | — | LiveCodeBench maintainers |
| OmniDocBench v1.6 | Multimodal | 89.1% | — | — | — | Linke Ouyang et al. |
| MMLU-Redux | Knowledge | 89.1% | 44.7 | — | — | Qwen |
| A Benchmark for Visual Question Answering using World Knowledge | Multimodal | 86.8% | — | — | — | Dustin Schwenk et al. |
| Graduate-Level Google-Proof Q&A | Knowledge | 85.8% | 57.4 | — | — | David Rein et al. |
| GPQA Diamond | Knowledge | 85.8% | 57.4 | — | — | David Rein et al. |
| τ²-Bench Tool-Agent-User Evaluation | Agentic | 80.2% | 56.4 | — | — | Victor Barres et al. |
| RealWorldQA | Multimodal | 80.1% | 46.7 | — | — | Qwen |
| CharXiv Descriptive and Reasoning Combined | Multimodal | 80.0% | — | — | — | CharXiv authors |
| Artificial Analysis Long Context Reasoning | Reasoning | 77.0% | 61.5 | — | — | Artificial Analysis |
| Berkeley Function Calling Leaderboard v3 | Agentic | 74.9% | — | — | — | Shishir G. Patil et al. |
| Instruction Following Benchmark | Instruction | 74.0% | 52.2 | — | — | Benchmark authors |
| Software Engineering Benchmark Verified | Coding | 60.8% | 46.3 | — | — | Carlos E. Jimenez et al. |
| BigCodeBench | Coding | 58.1% | — | — | — | DeepSeek-AI |
| OCRBench V2 | Multimodal | 56.9% | — | — | — | OCRBench authors |
| Terminal-Bench 2.1 (provider run) | Agentic | 52.8% | 55.2 | — | — | DeepSeek-AI |
| Terminal-Bench 2.1 (provider run) | Agentic | 52.8% | 55.2 | — | — | DeepSeek-AI |
13 benchmarks count, from 15 of 22 results. A grey row does not count. Too few models took that benchmark.