Quick answer: Llama 3.1 Nemotron Nano 8B v1 is NVIDIA's March 2025 Llama 3.1-based 8B model scoring 63.6% BFCL-v2 and 8.10/10 MT-Bench. Llama 3.1 License.
Where Nemotron Nano 8B v1 leads
Where it lags
Best for: NVIDIA-optimised Llama 3.1 8B inference; MT-Bench quality on NVIDIA hardware.
NVIDIA's Nemotron Nano 8B v1 is a fine-tuned version of Meta's Llama 3.1 8B, released March 2025 with NVIDIA's Nemotron post-training for better function calling and multi-turn quality. Superseded by the Nemotron Nano 9B v2 (June 2025).
| Field | Value |
|---|---|
| Organization | NVIDIA |
| License | Llama 3.1 License |
| HuggingFace | nvidia/Llama-3.1-Nemotron-Nano-8B-v1 |
| Release date | March 2025 |
| Parameters | 8B |
| Modality | Text only |
Open weights under Llama 3.1 License.
| Benchmark | Score | Source | Date |
|---|---|---|---|
| BFCL-v2 | 63.6% | Benchgen evaluation | 2025-03 |
| MT-Bench | 8.10/10 | Benchgen evaluation | 2025-03 |
| Model | BFCL | MT-Bench | License |
|---|---|---|---|
| Nemotron Nano 8B v1 | 63.6% | 8.10 | Llama 3.1 |
| Nemotron Nano 9B v2 | — | — | NVIDIA OML |
| Llama 3.1 8B Instruct | — | — | Llama 3.1 |
For current NVIDIA compact: use Nemotron Nano 9B v2 (Jun 2025) with better BFCL multi-turn (66.9%).
Specs from NVIDIA's Llama 3.1 Nemotron Nano 8B v1 release (March 2025) and Benchgen evaluations. Last updated 2026-07-24.
This model isn’t on any benchmark leaderboard yet.