Gemma-4-31B-IT-NVFP4 vs Qwen2.5-72B-Instruct
Specs, VRAM requirements and download trends — updated 6 August 2026.
Specification comparison
Differences are highlighted; identical values are muted.
| Specification | Gemma-4-31B-IT-NVFP4 | Qwen2.5-72B-Instruct |
|---|---|---|
| Deployment | Open weights | Open weights · also via API |
| Parameters | 20.9 B | 72.7 B |
| Architecture | Gemma4ForConditionalGeneration | Qwen2ForCausalLM |
| Context length | — | 32,768 |
| Licence | other | other |
| Commercial use | Unknown | Unknown |
| Languages | — | 1 |
| Downloads 30d | 1,136,331 | 166,748 |
| Downloads all time | 14.3 M | 9.1 M |
| Quantizations on the Hub | SAFETENSORS | SAFETENSORS |
| Gated | No | No |
| First seen on the Hub | 2026-04-02 | 2024-09-16 |
| API input price / 1M tokens | not on OpenRouter | $0.36 |
| API output price / 1M tokens | not on OpenRouter | $0.40 |
| Est. cost per 1M input + 1M output tokens | not on OpenRouter | $0.76 |
| Cached input / 1M tokens | not on OpenRouter | — |
| AA Intelligence Index | not on OpenRouter | — |
| AA Coding Index | not on OpenRouter | — |
| AA Agentic Index | not on OpenRouter | — |
| Max output tokens | not on OpenRouter | 16,384 |
| API providers | not on OpenRouter | 1 |
| Tool calling | not on OpenRouter | Yes |
| Reasoning mode | not on OpenRouter | No |
API cost
Estimated cost of 1M input + 1M output tokens via OpenRouter
Qwen2.5-72B-Instruct
$0.76
Download trend
Daily snapshots, last 90 days (13 Aug – 10 Nov)
2.4 M167 K
Gemma-4-31B-IT-NVFP4
Qwen2.5-72B-Instruct
VRAM side by side
On RTX 4090 · 24 GB · 8K context unless noted
Gemma-4-31B-IT-NVFP4 · fp16
39.5 GB / 24 GB
Qwen2.5-72B-Instruct · fp16
171.4 GB / 24 GB