LLM API cost calculator
Monthly API cost for your token volume across 35 models, at official list prices.
Lowest for this workload: Hy3, about $9.75 a month (75M tokens over 30 days).
| Model | Developer | Input / 1M | Cached / 1M | Output / 1M | Per request | Per month |
|---|---|---|---|---|---|---|
| Hy3 | Tencent | $0.080 | $0.0080 | $0.330 | $0.00032 | $9.75 |
| MiMo V2.6 Flash | Xiaomi | $0.140 | $0.0028 | $0.280 | $0.00042 | $12.6 |
| Claude Haiku 5.5 | Anthropic | $0.100 | $0.010 | $0.500 | $0.00045 | $13.5 |
| GLM 5.3 Flash (v3) | Zhipu | $0.150 | $0.030 | $0.500 | $0.00055 | $16.5 |
| GPT 5.6 Luna | OpenAI | $0.200 | $0.020 | $1.20 | $0.0010 | $30 |
| DeepSeek V4.1 Flash | DeepSeek | $0.300 | $0.0060 | $1.20 | $0.0012 | $36 |
| MiniMax M2.7 | MiniMax | $0.300 | $0.060 | $1.20 | $0.0012 | $36 |
| MiniMax M3 | MiniMax | $0.300 | $0.060 | $1.20 | $0.0012 | $36 |
| MiMo V2.6 Pro | Xiaomi | $0.435 | $0.0036 | $0.870 | $0.0013 | $39.1 |
| DeepSeek V4 Flash | DeepSeek | $0.440 | $0.014 | $1.32 | $0.0015 | $46.2 |
| Qwen3.7 Plus | Alibaba | $0.400 | $0.080 | $1.60 | $0.0016 | $48 |
| Composer 2.5 | Cursor | $0.500 | $0.200 | $2.50 | $0.0022 | $67.5 |
| Step 3.5 Flash | StepFun | $0.700 | $0.140 | $2.10 | $0.0024 | $73.5 |
| Hy4 Preview | Tencent | $0.834 | $0.042 | $2.50 | $0.0029 | $87.6 |
| Gemini 3.7 Flash | $0.750 | $0.075 | $3.75 | $0.0034 | $101 | |
| MiMo V2.5 Pro | Xiaomi | $1.00 | $0.200 | $3.00 | $0.0035 | $105 |
| Kimi K2.7 Code (v1) | Kimi | $0.950 | $0.190 | $4.00 | $0.0039 | $117 |
| DeepSeek V4 Pro | DeepSeek | $1.32 | $0.044 | $3.96 | $0.0046 | $139 |
| GLM 5.3 (v3) | Zhipu | $1.40 | $0.260 | $4.40 | $0.0050 | $150 |
| Step 3.7 Flash | StepFun | $1.35 | $0.270 | $8.10 | $0.0067 | $203 |
| Grok 4.5 | SpaceXAI | $2.00 | $0.300 | $6.00 | $0.0070 | $210 |
| Grok 4.6 | SpaceXAI | $2.00 | $0.500 | $6.00 | $0.0070 | $210 |
| Grok 4.7 | SpaceXAI | $2.00 | $0.500 | $6.00 | $0.0070 | $210 |
| Claude Sonnet 5 | Anthropic | $2.00 | $0.200 | $10 | $0.0090 | $270 |
| Claude Sonnet 5.5 | Anthropic | $2.00 | $0.100 | $10 | $0.0090 | $270 |
| GPT 5.6 Terra | OpenAI | $2.00 | $0.200 | $12 | $0.010 | $300 |
| MiMo V2.6 Pro UltraSpeed | Xiaomi | $4.35 | $0.036 | $8.70 | $0.013 | $392 |
| Kimi K3 (v1) | Kimi | $3.00 | $0.300 | $15 | $0.013 | $405 |
| Claude Opus 5.5 | Anthropic | $4.00 | $0.200 | $20 | $0.018 | $540 |
| GPT 5.6 Sol | OpenAI | $4.00 | $0.400 | $20 | $0.018 | $540 |
| Claude Opus 4.8 | Anthropic | $5.00 | $0.500 | $25 | $0.022 | $675 |
| Claude Opus 5 | Anthropic | $5.00 | $0.500 | $25 | $0.022 | $675 |
| Step 5 Preview | StepFun | $7.00 | $0.350 | $20 | $0.024 | $720 |
| GPT 5.5 | OpenAI | $5.00 | $0.500 | $30 | $0.025 | $750 |
| Claude Fable 5 | Anthropic | $10 | $1.00 | $50 | $0.045 | $1,350 |
Estimates use standard list prices in USD and a 30-day month. Batch and off-peak discounts, cache-write fees, tool calls and taxes are left out. Where no cached price is listed, cached tokens are charged at the input price.
Data: real-api-pricing (MIT) · prices as of Oct 16, 2026
How the estimate works
Cost per request = fresh input tokens × input price + cached input tokens × cached price + output tokens × output price, with every price per 1 million tokens. The cached share splits your input tokens between the two input prices.
Monthly cost = cost per request × requests per day × 30. Output tokens are usually the most expensive part, so long answers move the total more than long prompts.
Questions
- How do I calculate LLM API cost?
- Multiply input tokens by the input price and output tokens by the output price, each per 1 million tokens, then add them. Cached input is billed at a lower cached price on most APIs. Multiply by requests per month for a monthly total.
- Which model has the cheapest API price?
- Of the 35 models listed here, Hy3 has the lowest input price at $0.080 per 1M tokens and MiMo V2.6 Flash has the lowest output price at $0.280 per 1M tokens (prices as of Oct 16, 2026).
- What does 1,000 requests a day cost?
- At 2,000 input and 500 output tokens per request with no caching, the lowest monthly cost here is Hy3 at about $9.75.
- Which model is the most expensive per token?
- Claude Fable 5 has the highest combined list price here: $10 per 1M input tokens and $50 per 1M output tokens.
- How much does prompt caching save?
- Cached input usually costs a fraction of fresh input. Set the cached share to see the effect for your workload; some vendors also charge for writing to the cache, which this estimate leaves out.
- What does this estimate leave out?
- Batch and off-peak discounts, cache-write fees, tool and search calls, images and audio, and taxes. Check the vendor's pricing page before you budget.