UberLLM lets you call GPT models at the cheapest healthy seller, or drop to much cheaper open-weight models by changing the model id — all with the OpenAI SDK and one key. UberLLM gives you one OpenAI-compatible API key and base URL for every major model, routing every request to the cheapest healthy seller for that model and billing the price shown, with UberLLM's 10% fee already included.
| Model | Input / 1M | Output / 1M |
|---|---|---|
| GPT-OSS 20B | $0.0002 | $0.0009 |
| GPT-OSS 120B | $0.0007 | $0.0029 |
| GPT OSS Safeguard 20B | $0.0008 | $0.0030 |
| GPT-6 Luna Decisions (OpenAI) | $0.0220 | $0.0000 |
| GPT OSS Safeguard 120B | $0.0028 | $0.0112 |
| GPT-5.6 Luna | $0.0022 | $0.0133 |
| GPT-6 Luna | $0.0033 | $0.0164 |
| GPT-4o Mini | $0.0049 | $0.0197 |
| GPT 5 Nano | $0.0066 | $0.0528 |
| GPT OSS 20B | $0.0138 | $0.0522 |
| GPT 5 Mini | $0.0165 | $0.1320 |
| GPT-5.3 Codex | $0.0211 | $0.1685 |
Live prices via UberLLM · full pricing table →
UberLLM routes GPT to the cheapest healthy seller and makes it trivial to switch to cheaper models (DeepSeek, GLM, Qwen) for cost-sensitive workloads — same SDK, one key.