If you're processing lots of text (summaries, classification, extraction, chatbots), the price per million tokens matters more than anything else. Here are the cheapest chat models on Solo Router today.
Cheapest chat models
| Model | Input / 1M | Output / 1M | Context |
|---|---|---|---|
| DeepSeek V4 Flash | $0.18 | $0.36 | 1M |
| GLM-5.3 Flash | $0.20 | $0.65 | 1M |
| MiniMax M3.1 Flash | $0.20 | $0.78 | 1M |
| Gemma 4 31B | $0.23 | $0.65 | 256K |
| Ministral 3 14B | $0.26 | $0.26 | 262K |
| Nemotron Nano 12B v2 VL | $0.26 | $0.78 | 128K |
Prices update automatically from our catalog. See every model on the Models page.
Is a cheap model good enough?
Usually, yes, for:
- summarising, tagging and classifying text;
- extracting fields into JSON;
- customer-support style chat with a good system prompt;
- first drafts and rewrites.
Use a bigger model (GLM-5.3, Kimi K3, Qwen 3.8 Max, DeepSeek V4 Pro) for:
- multi-step coding agents;
- hard reasoning and maths;
- long documents where details matter.
A smart pattern: cheap first, strong when needed
Because every model sits behind the same API key on Solo Router, you can route most requests to a cheap model and send only the hard ones to a strong model. Just change the model string. No second account, no second bill.
Save more with caching and limits
- Keep the start of your prompt identical between calls so it's billed at the cached price.
- Set
max_tokensto cap output length. - Give each key a spending limit on the Keys page.