head-to-head

NVIDIA: Nemotron 3.5 Super VL vs PrismML: Ternary Bonsai 2 27B

Side-by-side comparison of specs, pricing, benchmark scores, and task rankings. Updated 2026-10-19.

The verdict

For a sample job of 1 million input tokens and 250,000 output tokens, both cost $0.20 at current list prices.

Nemotron 3.5 Super VL reads up to 1,048,576 tokens at once, roughly 1,573 pages of text, against about 393 pages for Ternary Bonsai 2 27B.

Only Nemotron 3.5 Super VL offers video input.

Across the 18 tasks where they differ, Nemotron 3.5 Super VL scores higher on 18 (SQL Generation, Code Review, Code Completion and others).

Nemotron 3.5 Super VL was released in October 2026 and Ternary Bonsai 2 27B in September 2026.

Pick Nemotron 3.5 Super VL if you need long documents, video input or the stronger all-round task scores.

Costs use current list prices per million tokens; page counts assume about 0.75 words per token and 500 words per page.

Task	NVIDIA: Nemotron 3.5 Super VL	PrismML: Ternary Bonsai 2 27B
SQL Generation	139	131
Code Review	140	126
Code Completion	132	129
Code Refactoring	143	127
Bug Fixing	144	130
Unit Test Generation	129	121
Code Documentation	135	126
Regex Writing	124	119
CI/CD Pipelines	125	117
Frontend Component Design	127	122
Data Analysis	130	124
CSV / Spreadsheet Cleanup	135	128
ETL Scripting	134	122
OCR / Document Parsing	133	128
Table Extraction from PDFs	133	128
Long-Document Summarization	144	129
Short-Form Summarization	126	123
Blog Post Writing	127	119

Scores reflect capability match + benchmark data + pricing for each task. Methodology →