Solar Open 100B
Solar Open 100B is an open-weights language model from Upstage, released December 31, 2025. It has 103B parameters (mixture-of-experts, 12B active). The context window is 128K tokens.
Solar Open is Upstage's flagship 102B-parameter large language model, trained entirely from scratch and released under the Solar-Apache License 2.0 (see LICENSE).
Check it against your own PC →
Can you run Solar Open 100B locally?
The smallest common GPU it runs on is 6 GB of VRAM (with 32 GB of RAM): IQ2_XXS, split GPU / CPU, roughly 10 tokens per second. Estimates assume 32 GB of DDR5 RAM, a 16K context and llama.cpp-style GGUF quantization. Enter your exact hardware in the live app.
| GPU memory | Verdict | Best quant | VRAM | RAM | tok/s |
|---|---|---|---|---|---|
| 6 GB VRAM | Runs, slowly | IQ2_XXS (heavy loss) | 5.2 GB | 25.1 GB | ~10 |
| 8 GB VRAM | Runs · GPU + CPU experts | IQ2_XXS (heavy loss) | 7.2 GB | 23.1 GB | ~16 |
| 12 GB VRAM | Runs · GPU + CPU experts | IQ2_XXS (heavy loss) | 11.2 GB | 19.1 GB | ~19 |
| 16 GB VRAM | Runs · GPU + CPU experts | IQ2_M (heavy loss) | 15.2 GB | 23.3 GB | ~17 |
| 24 GB VRAM | Runs · GPU + CPU experts | IQ3_XXS (noticeable loss) | 23.2 GB | 19.9 GB | ~20 |
| 32 GB VRAM | Runs · GPU + CPU experts | Q3_K_M (okay) | 31.2 GB | 22.8 GB | ~18 |
Ready-made quantizations on Hugging Face: mradermacher/Solar-Open-100B-GGUF · mradermacher/Solar-Open-100B-i1-GGUF · AaryanK/Solar-Open-100B-GGUF · DevQuasar-11/upstage.Solar-Open-100B-GGUF · cyankiwi/Solar-Open-100B-AWQ-4bit · cyankiwi/Solar-Open-100B-AWQ-8bit
Benchmark results
Epoch AI has not published results for Solar Open 100B yet. New models are usually evaluated within days to weeks.
Links
Announcement · Hugging Face: upstage/Solar-Open-100B
More from Upstage
Solar Mini 4 · Solar Pro 4 · Solar Open2 250B · solar-pro3 · solar-pro2