Rank #43
PrismML: Ternary Bonsai 2 9B
- Multimodal
- 262k ctx
- 1M
- $0.11
Model · Vendors
PrismML
Ranked models from this vendor or team on Global rankings (same data snapshot as the main global board).
Board snapshot:
About this vendor
PrismML publishes ternary-compressed open-weight models such as Ternary Bonsai 2 27B (derived from Qwen3.8-27B) on aggregator catalogs including OpenRouter. On AI Hippo this vendor hub groups PrismML-tagged SKUs with Global Rankings placement, context metrics, and snapshot token pricing.
Vendor snapshot
Catalog models 2
Ranked on Global board 1
Best rank #43
Price range (avg/1M) $0.11–$0.29/1M
Max context 262k ctx
Model types
Top ranked models
Models on this board · Global rankings
| Rank | Name | Type | Key metric | 1M tokens (avg) |
|---|---|---|---|---|
| 43 | PrismML: Ternary Bonsai 2 9B | Multimodal | 262k ctx | $0.11 |
Catalog models (not in ranking)
| Rank | Name | Size | Price | Notes |
|---|---|---|---|---|
| — | PrismML: Ternary Bonsai 2 27B | 262k ctx | $0.29/1M | prism-ml/ternary-bonsai-2-27b |
| — | PrismML: Ternary Bonsai 2 9B | 262k ctx | $0.11/1M | prism-ml/ternary-bonsai-2-9b |
Data sources & methodology
Counts and prices come from the OpenRouter catalog snapshot merged with AI Hippo Global rankings (same snapshot as /rankings/). Rank reflects the composite score on that board—not LMSYS Chatbot Arena ELO. Snapshot date: 2026-11-02. Methodology
FAQ
What is PrismML's highest-ranked model?
PrismML: Ternary Bonsai 2 9B currently ranks #43 on AI Hippo's Global rankings board (snapshot 2026-11-02).
What does PrismML charge per 1M tokens?
Across catalog and ranked rows, blended 1M-token prices range from $0.11 to $0.29 in the latest snapshot (2026-11-02). Check individual model pages for prompt vs completion splits.
How many PrismML models are ranked vs catalog-only?
1 of 2 catalog models appear on Global rankings; the remaining 1 are listed under catalog-only on this page.
How is the Global rankings score calculated?
Global rankings uses on-site weighting (context window first, then blended 1M-token price)—not third-party ELO. Full methodology: /en/methodology/