The cheapest LLM API is the one that always routes you to the lowest-cost healthy seller. UberLLM does exactly that, for every request: small open-weight models like DeepSeek, GLM and Qwen cost only a few cents per million tokens, and you reach all of them — plus GPT, Claude and Gemini — through a single OpenAI-compatible key. Prices include UberLLM's 10% fee, so the price you see is the price you pay.
| Model | Input / 1M | Output / 1M |
|---|---|---|
| JEV 1.13 (TypeSafe) | $0.0005 | $0.0000 |
| Qwen: Qwen3.8 Omni Flash | $0.0002 | $0.0005 |
| Gemma 3 4B IT | $0.0005 | $0.0007 |
| GPT-OSS 20B | $0.0002 | $0.0009 |
| DeepSeek V4.1 Flash | $0.0004 | $0.0009 |
| Ministral 3B | $0.0011 | $0.0011 |
| MiniMax M2.1 | $0.0004 | $0.0016 |
| Gemma 4 26B A4B IT | $0.0005 | $0.0016 |
| Gemma 3 12B IT | $0.0005 | $0.0016 |
| DeepSeek V4 Flash 0731 | $0.0010 | $0.0020 |
| GLM 4.7 Flash | $0.0003 | $0.0022 |
| Gemma 3 27B | $0.0009 | $0.0022 |
Live prices via UberLLM · full pricing table →
For most workloads, small open-weight models (DeepSeek, GLM-4.x, Qwen) served through UberLLM are the cheapest — often a few cents per million tokens. UberLLM routes to the cheapest healthy seller automatically.
Yes — via UberLLM you can call GPT models through the cheapest healthy seller, or switch to much cheaper open-weight models with one line of code (change the model id). Same OpenAI SDK.