Leaderboard / LumiOpen / AMD Silo AI (Poro)
Poro 2 70B Instruct
LumiOpen / AMD Silo AI (Poro) · Poro · released 2025-05-27 · data checked 2026-11-01
Poro 2 70B Instruct is a 70.6B-parameter dense model from LumiOpen / AMD Silo AI (Poro) (Europe), released 2025-05-27 under the Llama 3.3 Community license. It doesn't have enough reported benchmarks for an index score yet. It needs about 47.3 GB of memory at 4-bit with an 8K-token context.
Llama 3.1 70B with continued pre-training for Finnish.
Benchmarks
Each bar shows this model's score. The tick marks the best large model on that benchmark. Sources: 1 reported by LumiOpen / AMD Silo AI (Poro). How we label sources.
Specifications
- Lab
- LumiOpen / AMD Silo AI (Poro) (FI · Europe)
- Based on
- Llama 3.1 70B (fine-tune or distillation)
- Architecture
- Dense transformer
- Total parameters
- 70.6B
- Active parameters
- All (dense)
- Size class
- Large
- Context window
- 8K tokens
- Input modalities
- text
- Reasoning
- Instruct (no thinking)
- License
- Llama 3.3 Community
- Openness
- Open weights
- Memory to run
- ~47.3 GB at 4-bit, 8K context (more sizes)
- Hugging Face
- 1.6K downloads (30 days) · 9 likes
Compare with
- Poro 2 70B Instruct vs DeepSeek-R1-Distill-Llama-70B (70.6B, index 50.1)
- Poro 2 70B Instruct vs Llama 3.3 70B (70.6B, index 40.8)
- Poro 2 70B Instruct vs K2-V2 Instruct (70B, index 57.9)
Sources
Something wrong here? Report it.
Run it locally
No GGUF build from a well-known publisher yet.
vllm serve LumiOpen/Llama-Poro-2-70B-InstructMemory needed: weights, context cache and runtime buffers.
| Quantization | 4K context | 8K context |
|---|---|---|
| 16-bit F16 | 144 GB | 146 GB |
| 8-bit Q8_0 | 78 GB | 80 GB |
| 4-bit Q4_K_M | 46 GB | 47 GB |