The honest, $0-markup cost to run gpt-oss-120b (OpenAI, 117B) and Qwen3 Coder 30B A3B Instruct (Qwen, 30.5B), side by side — free on your own machine, on your own rented GPU at the vendor's price, or via your own API key.
| gpt-oss-120b | Qwen3 Coder 30B A3B Instruct | |
|---|---|---|
| Parameters | 117B | 30.5B |
| Context window | 131.1K | 262.1K |
| License | Commercial OK | Commercial OK |
| VRAM to run | ~80 GB (Q4_K_M) | ~22 GB (Q4_K_M) |
| Rent a GPU | $0.98/hr | $0.26/hr |
| Your API key | $0.11/1M | $0.34/1M (est.) |
VRAM, size, context and license are facts from the catalog and the shared cost engine; API prices are real where we have them (labeled "est." otherwise). We don't rank model quality or quote benchmarks here.
Full cost breakdown: gpt-oss-120b →
Full cost breakdown: Qwen3 Coder 30B A3B Instruct →
Open the free Spanvero advisor → to compare them live for your exact workload and hardware.
Spanvero · All comparisons · Prices as of 2026-06-17. $0 markup, your own accounts, we never resell compute. © 2026 Cynosure LLC.