Bonsai 2 2B vs Bonsai 2 4B
Side-by-side VRAM requirements, benchmark scores, and GPU compatibility for local AI inference.
Quick verdict
Bonsai 2 2B is more hardware-efficient: it needs 0.7 GB at its Ternary (PQ2_0) build vs 1.6 GB for Bonsai 2 4B's Ternary (PQ2_0), fitting on 117 GPUs natively.
VRAM at each real build (8k context)
Bonsai 2 2B and Bonsai 2 4B both ship as a handful of fixed prebuilt files rather than a standard quantization ladder, so the two don't share equivalent bit-widths to line up row for row. Ranked smallest to largest instead.
| Rank | Bonsai 2 2B | GB | Bonsai 2 4B | GB |
|---|---|---|---|---|
| 1 | Ternary (PTQ1_0) | 0.6 GB | Ternary (PTQ1_0) | 1.4 GB |
| 2 | Ternary (PQ2_0) | 0.7 GB | Ternary (PQ2_0) | 1.6 GB |
| 3 | Ternary (Q2_0) | 0.7 GB | Ternary (Q2_0) | 1.7 GB |
Each column is that model's own real builds, smallest first; rank pairs them by position, not by matching precision.
Model specifications
| Spec | Bonsai 2 2B | Bonsai 2 4B |
|---|---|---|
| Org | PrismML | PrismML |
| Parameters | 2B | 4B |
| Architecture | Dense | Dense |
| Context | 256k tokens | 256k tokens |
| Modalities | text, vision | text, vision |
| License | Apache 2.0 | Apache 2.0 |
| Commercial | Yes | Yes |
| Released | 2026-10-15 | 2026-10-15 |
| GPUs (native) | 117 / 121 | 117 / 121 |
Benchmark scores
| Benchmark | Bonsai 2 2B | Bonsai 2 4B |
|---|---|---|
| GPQA Diamond | 39.6 | 63.8 |
| IFEval | 85.8 | 88.9 |
Green = higher score (better). N/A = not yet available. ~ = inherited from a base model, not independently reported for that release itself.
GPUs that run only Bonsai 2 2B(0)
Every GPU that runs Bonsai 2 2B also runs Bonsai 2 4B.
GPUs that run only Bonsai 2 4B(0)
Every GPU that runs Bonsai 2 4B also runs Bonsai 2 2B.
GPUs that run both natively(117)
- NVIDIA RTX 509032 GB
- NVIDIA RTX 508016 GB
- NVIDIA RTX 5070 Ti16 GB
- NVIDIA RTX 507012 GB
- NVIDIA RTX 5060 Ti 16GB16 GB
- NVIDIA RTX 5060 Ti 8GB8 GB
- NVIDIA RTX 50608 GB
- NVIDIA RTX 50508 GB
- NVIDIA RTX 409024 GB
- NVIDIA RTX 408016 GB
- NVIDIA RTX 4070 Ti SUPER16 GB
- NVIDIA RTX 4070 Ti12 GB
- +105 more GPUs run both
Which should you use?
- • You have limited VRAM: it's a smaller model needing 0.7 GB vs 1.6 GB
- • You want maximum capability and have a 2 GB+ GPU
Frequently asked questions
- Which is better, Bonsai 2 2B or Bonsai 2 4B?
- Bonsai 2 2B has 2B parameters vs 4B for Bonsai 2 4B, so Bonsai 2 4B is the larger model. Bonsai 2 2B is more hardware-efficient, needing 0.7 GB at its Ternary (PQ2_0) build vs 1.6 GB for Bonsai 2 4B's Ternary (PQ2_0).
- How much VRAM does Bonsai 2 2B need vs Bonsai 2 4B?
- At 8k context, Bonsai 2 2B needs approximately 0.7 GB of VRAM at its Ternary (PQ2_0) build, while Bonsai 2 4B needs 1.6 GB at its Ternary (PQ2_0) build. At the largest build each ships, Bonsai 2 2B requires 0.7 GB (Ternary (Q2_0)) vs 1.7 GB (Ternary (Q2_0)) for Bonsai 2 4B.
- Can you run Bonsai 2 2B on the same GPUs as Bonsai 2 4B?
- Yes, 117 GPUs can run both natively in VRAM, including NVIDIA RTX 5090, NVIDIA RTX 5080, NVIDIA RTX 5070 Ti. However, no GPU can run Bonsai 2 2B without also fitting Bonsai 2 4B, and no GPU can run Bonsai 2 4B without also fitting Bonsai 2 2B.
- What is the difference between Bonsai 2 2B and Bonsai 2 4B?
- Bonsai 2 2B has 2B parameters (dense) with a 256k context window. Bonsai 2 4B has 4B parameters (dense) with a 256k context window.
- Which model fits in 24 GB of VRAM, Bonsai 2 2B or Bonsai 2 4B?
- Both fit in 24 GB of VRAM at their respective recommended builds: Bonsai 2 2B (Ternary (PQ2_0)) needs 0.7 GB and Bonsai 2 4B (Ternary (PQ2_0)) needs 1.6 GB.