DeepSeek V3 Pricing
The most popular open-source LLM. Prices per 1 million tokens.
| Provider | Input Price | Output Price | Total (50/50 mix) | vs NovAI |
|---|---|---|---|---|
| NovAI (HK) | $0.20 | $0.40 | $0.30 | — |
| DeepSeek Official | $0.28 | $0.42 | $0.35 | NovAI 17% cheaper |
| Novita.ai | $0.27 | $0.40 | $0.34 | NovAI 12% cheaper |
| SiliconFlow | $0.27 | $0.42 | $0.35 | NovAI 14% cheaper |
| OpenRouter | $0.32 | $0.89 | $0.61 | NovAI 51% cheaper |
| Fireworks.ai | $0.56 | $1.68 | $1.12 | NovAI 73% cheaper |
| Together.ai | $0.60 | $1.70 | $1.15 | NovAI 74% cheaper |
GLM Model Pricing
Zhipu AI vision-language models. Prices per 1 million tokens.
| Provider | Model | Input Price | Output Price | Notes |
|---|---|---|---|---|
| NovAI | GLM-4.6V-Flash | FREE | FREE | Unique — no other provider offers this free |
| NovAI | GLM-4.6V | $0.40 | $1.20 | 33% cheaper output than competitors |
| SiliconFlow | GLM-4.5-Air | $0.14 | $0.86 | Different model version |
| Together.ai | GLM-4.6 | $0.60 | $2.20 | |
| Fireworks.ai | GLM-4.7 | $0.60 | $2.20 | |
| Novita.ai | GLM-4.7 | $0.60 | $2.20 |
MiniMax Pricing
MiniMax large language models. Prices per 1 million tokens.
| Provider | Input Price | Output Price | Context Window |
|---|---|---|---|
| NovAI | $0.20 | $1.60 | 1M tokens |
| Together.ai | $0.30 | $1.20 | 1M tokens |
| Fireworks.ai | $0.30 | $1.20 | 1M tokens |
| Novita.ai | $0.30 | $1.20 | 1M tokens |
Why is NovAI cheaper?
NovAI is a lean operation based in Hong Kong with direct peering to Chinese AI providers. We have lower infrastructure overhead than US-based competitors, and we pass those savings to developers. Our servers are one network hop from DeepSeek and Zhipu AI data centers, which also means lower latency (~80ms vs 300ms+ from the US). We accept USDT (TRC20) payments — no credit card processing fees.
Methodology
Prices were collected from each provider's official pricing page in March 2026. For providers with multiple tiers, we used the standard/on-demand pricing (no volume discounts). The "Total (50/50 mix)" column assumes an equal split between input and output tokens. Your actual costs will depend on your specific input/output ratio. Prices may change — always check the provider's official page for the latest rates.
Start with the cheapest DeepSeek API
Sign up in 30 seconds. No credit card required. Try GLM-4.6V-Flash completely free.
Get Started Free →