Measured over the last 30 days
- Models Tracked
- 59
- Avg Tokens / Second
- 17.92
- Avg Time to First Token (ms)
- 9751.86
- Last Updated
- Oct 14, 2026
All novita models
59 live · one row per model · fastest first · 3 retired lanes not listed · 6 folded behind their current lane| Model | Avg Toks/Sec | Min | Max | Avg TTF (ms) |
|---|---|---|---|---|
| Nemotron 3 Nano 30B A3B | 56.50 | 56.50 | 56.50 | 760.00 |
| Llama 3 8B Lunaris | 52.90 | 48.90 | 56.90 | 550.00 |
| Llama 3.1 8B Instruct | 51.30 | 51.30 | 51.30 | 920.00 |
| Mistral Nemo | 46.20 | 46.20 | 46.20 | 710.00 |
| Ling 3.0 Flash | 42.80 | 42.80 | 42.80 | 1330.00 |
| Qwen3 VL 30B A3B Instruct | 41.90 | 34.00 | 49.70 | 870.00 |
| Qwen3 Next 80B A3B Instruct | 39.90 | 39.90 | 39.90 | 1300.00 |
| Qwen3 Coder Next | 35.20 | 35.20 | 35.20 | 1430.00 |
| Gemma 4 26B A4B | 35.10 | 3.33 | 58.70 | 2250.00 |
| Qwen3 Coder 30B A3B Instruct | 33.70 | 33.70 | 33.70 | 1370.00 |
| Llama 3.1 Euryale 70B v2.2 | 33.10 | 33.10 | 33.10 | 650.00 |
| Qwen3 VL 235B A22B Instruct | 32.80 | 32.80 | 32.80 | 950.00 |
| Llama 4 Maverick | 31.40 | 17.70 | 45.00 | 1470.00 |
| Kimi K2 0711 | 27.70 | 23.00 | 32.40 | 1010.00 |
| Llama 3.3 70B Instruct | 27.40 | 22.40 | 32.70 | 1000.00 |
| Llama 4 Scout | 26.00 | 26.00 | 26.00 | 1160.00 |
| gpt-oss-20b | 24.50 | 8.70 | 55.00 | 2810.00 |
| Kimi K2 0905 | 23.30 | 19.10 | 27.50 | 1440.00 |
| gpt-oss-120b | 22.50 | 3.10 | 33.10 | 7270.00 |
| DeepSeek V3.1 Terminus | 21.70 | 21.70 | 21.70 | 1210.00 |
| Qwen3 235B A22B Thinking 2507 | 20.80 | 2.79 | 38.90 | 11500.00 |
| MiniMax M3 | 19.80 | 6.41 | 51.60 | 3800.00 |
| DeepSeek V3.2 | 19.60 | 9.78 | 24.80 | 1770.00 |
| MiMo-V2.5-Pro | 19.00 | 19.00 | 19.00 | 2390.00 |
| GLM 5.2 | 18.40 | 14.40 | 23.20 | 2600.00 |
| Qwen3 235B A22B Instruct 2507 | 16.30 | 7.26 | 27.10 | 1290.00 |
| DeepSeek V3.2 Exp | 15.90 | 15.90 | 15.90 | 1490.00 |
| MiniMax M1 | 15.00 | 15.00 | 15.00 | 5200.00 |
| DeepSeek V4 Flash 0731 | 14.80 | 2.87 | 24.00 | 4630.00 |
| Gemma 4 31B | 14.60 | 12.60 | 17.30 | 1160.00 |
| WizardLM-2 8x22B | 14.60 | 14.60 | 14.60 | 710.00 |
| MiMo-V2.5 | 13.90 | 13.90 | 13.90 | 2810.00 |
| GLM 4.5 Air | 12.10 | 10.20 | 13.40 | 4650.00 |
| Ling 3.1 Flash | 10.90 | 7.61 | 13.80 | 5220.00 |
| Kimi K2.7 Code | 10.90 | 7.09 | 17.00 | 5160.00 |
| GLM 5.1 | 10.90 | 1.85 | 20.00 | 17600.00 |
| DeepSeek V4 Flash 0423 | 9.44 | 3.94 | 23.10 | 7650.00 |
| Gemma 3 27B | 9.36 | 9.36 | 9.36 | 1390.00 |
| MiniMax M2.1 | 9.05 | 9.05 | 9.05 | 8420.00 |
| Qwen3.5-122B-A10B | 7.61 | 7.61 | 7.61 | 8010.00 |
| Qwen3 Coder 480B A35B | 7.54 | 7.54 | 7.54 | 6640.00 |
| MiMo-V2.6-Pro | 7.46 | 4.56 | 9.91 | 6710.00 |
| Qwen3.5-27B | 5.25 | 5.25 | 5.25 | 11500.00 |
| MiniMax M2 | 4.88 | 4.88 | 4.88 | 13600.00 |
| MiniMax M2.7 | 4.56 | 2.94 | 6.91 | 17800.00 |
| Kimi K2.6 | 4.50 | 2.54 | 6.70 | 14400.00 |
| MiniMax M2.5 | 4.15 | 3.90 | 4.41 | 15300.00 |
| Kimi K2.5 | 4.03 | 3.75 | 4.31 | 14400.00 |
| Kimi K2 Thinking | 3.36 | 2.86 | 3.86 | 18300.00 |
| Qwen3.5 397B A17B | 3.31 | 2.52 | 4.46 | 19300.00 |
| R1 | 3.27 | 2.76 | 3.79 | 17400.00 |
| Step 5 Preview | 3.04 | 3.04 | 3.04 | 19800.00 |
| GLM 5 | 2.64 | 2.64 | 2.64 | 23200.00 |
| GLM 4.5V | 2.45 | 2.45 | 2.45 | 24700.00 |
| Qwen3.8 27B | 1.98 | 0.97 | 3.00 | 38500.00 |
| GLM 4.6 | 1.91 | 1.45 | 2.37 | 33800.00 |
| GLM 4.7 | 1.45 | 1.39 | 1.51 | 42200.00 |
| MiMo-V2.6-Flash | 1.43 | 1.43 | 1.43 | 45200.00 |
| GLM 4.6V | 0.96 | 0.96 | 0.96 | 64700.00 |
Frequently Asked Questions
Which novita model is fastest?
Based on recent tests, Nemotron 3 Nano 30B A3B shows the highest average throughput among tracked novita models.
How many recent measurements feed this dashboard?
This provider summary aggregates 3648 individual prompts measured across 3083 monitoring runs over the past month.