Modelverse keeps model facts separate from Run evidence and warns whenever benchmark conditions are not equivalent.
These models do not currently have leading Runs under identical evaluation conditions. Values are shown with their provenance and should not be treated as a universal leaderboard.
| Attribute | Sparrow Chat 3B local-ai/sparrow-chat-3b | Nova Instruct 12B community/nova-instruct-12b |
|---|---|---|
| Openness | Open weights | Open weights |
| Commercial use | Permitted | Permitted |
| Parameters | 3.2B | 12B |
| Context length | 16K tokens | 33K tokens |
| Estimated memory | 2.8 GB · Q4_K_M | 8.2 GB · Q4_K_M |
| Runtimes | Ollama, llama.cpp, MLX | Ollama, llama.cpp, Transformers, vLLM |
| Tasks | Chat, Summarization, Extraction | Chat, Summarization, Reasoning, Translation |
| Verified Runs | 16 | 9 |
| Best evidence | Independently reproduced | Machine-captured |
| Task score | — | 71.3 Multilingual instruction suite |
| Speed | 18.9 tok/s | 118.4 tok/s |
| Peak memory | 3.5 GB | 18.8 GB |
| Run hardware | Apple M1 · 8 GB | RTX 3090 · 24 GB |
A fast, memory-efficient chat model for laptops and edge devices.
A general-purpose multilingual instruction model with broad runtime support.
First: eliminate models that fail the license, task, modality or runtime requirement.
Second: confirm fit on equivalent hardware or preserve a generous estimated memory margin.
Third: compare only Runs sharing an evaluation suite, revision and broadly equivalent configuration.
