Nanbeige4.3-33B
Nanbeige · Nanbeige4.3
33.0B
4 days ago
118 open models tracked live
What can you run on your GPU?
Pick your graphics card. See which open-source LLMs fit, how fast they run, and how much VRAM they need.
118
Open Models
391
Benchmarks Tracked
90
GPUs Compared
Top Open-Source Models
Ranked by composite benchmark score & live trend momentum
1
Nanbeige4.3-33B
25644.5
composite score
2
Raon-1 mini
8.4
composite score
3
Rising
Qwen3.8-27B
2.6
composite score
43%
| # | Model | Score | VRAM (Q4) | Trend | License | Compare |
|---|---|---|---|---|---|---|
| 4 |
Muse-Spark-Open-816B-A42B
Meta Superintelligence Lab · 816.0B |
2.6 | 408GB | — | Muse Model License | |
| 5 |
Gemma4-26B-A4B
Google · 25.2B |
2.5 | 13GB | Steady | Apache License 2.0 | |
| 6 |
DeepSeek-V4.1-Flash-1017
DeepSeek · 763.0B |
2.4 | 382GB | — | MIT License |
Latest Releases
Newest open-source models, freshest first
GPU Price Comparison
Cheapest EU hosting rates for LLM inference — live via GPU Radar
Best value right now
GeForce RTX 3090 — $0.21/h
24GB VRAM · 936 GB/s · Cloud · $0.009/GB·h
| GPU | VRAM | Bandwidth | Cheapest /h | /GB·h | Best For |
|---|---|---|---|---|---|
| GeForce RTX 3090 | 24GB | 936 GB/s | $0.21/h Partner | $0.009 | enthusiast |
| GeForce RTX 4090 | 24GB | 1008 GB/s | $0.4/h Partner | $0.017 | flagship |
| Apple M1 | 8GB | 68 GB/s | $0.65/h Partner | $0.081 | entry |
| Apple M2 | 8GB | 100 GB/s | $0.65/h Partner | $0.081 | entry |
| Apple M4 | 16GB | 120 GB/s | $0.65/h Partner | $0.041 | entry |
| Apple M3 | 8GB | 100 GB/s | $0.65/h Partner | $0.081 | entry |
EU hosting prices via GPU Radar (Hetzner, Vast.ai, RunPod and more); USD cloud rates as fallback. Links may be affiliate or referral links.
Compare Models Head-to-Head
Benchmark scores side by side
Popular comparisons
License Quick-Check
Find models you can actually use commercially
Trending Models
Development across downloads, likes & followers — ranked by momentum