Pay-as-you-go API prices
Pay-as-you-go API prices
Pay-as-you-go list prices per 1M tokens. Cache reads and output dominate the blended cost, so the ranking follows your cache hit rate rather than the headline input price.
Pay-as-you-go API prices
Pay-as-you-go list prices per 1M tokens, ranked by what 1M mixed tokens cost at your settings. Cache read and output prices decide the ranking far more than the headline input price.
| Model | Provider | Input | Cached read | Output | Blended $/1M |
|---|---|---|---|---|---|
| muse-spark-1.2-contributor | Meta | $0.100 | $0.0020 | $0.200 | $0.0098 |
| muse-spark-1.3-contributor | Meta | $0.100 | $0.0020 | $0.200 | $0.0098 |
| ministral-3b-latest | Mistral | $0.100 | $0.010 | $0.100 | $0.014 |
| gpt-5-nano | OpenAI | $0.050 | $0.0050 | $0.400 | $0.018 |
| devstral-small-latest | Mistral | $0.100 | $0.010 | $0.300 | $0.020 |
| mistral-small | Mistral | $0.100 | $0.010 | $0.300 | $0.020 |
| ministral-8b-latest | Mistral | $0.150 | $0.015 | $0.150 | $0.022 |
| gemini-2.5-flash-lite | $0.100 | $0.010 | $0.400 | $0.023 | |
| voxtral-small-latest | Mistral | $0.100 | $0.010 | $0.400 | $0.023 |
| deepseek-coder | DeepSeek | $0.140 | $0.014 | $0.280 | $0.024 |
| claude-haiku-5-5 | Anthropic | $0.100 | $0.010 | $0.500 | $0.027 |
| gpt-6-luna | OpenAI | $0.100 | $0.010 | $0.500 | $0.027 |
| ministral-14b-latest | Mistral | $0.200 | $0.020 | $0.200 | $0.029 |
| qwen3.8-omni-flash | Alibaba Qwen | $0.150 | $0.016 | $0.470 | $0.032 |
| qwen3.8-flash | Alibaba Qwen | $0.150 | $0.016 | $0.470 | $0.033 |
| magistral-small-latest | Mistral | $0.150 | $0.015 | $0.600 | $0.035 |
| mistral-small-latest | Mistral | $0.150 | $0.015 | $0.600 | $0.035 |
| mistral-vibe-cli-fast | Mistral | $0.150 | $0.015 | $0.600 | $0.035 |
| codestral-mamba-latest | Mistral | $0.250 | $0.025 | $0.250 | $0.036 |
| mistral-tiny | Mistral | $0.250 | $0.025 | $0.250 | $0.036 |
| open-mistral-nemo | Mistral | $0.300 | $0.030 | $0.300 | $0.043 |
| deepseek-v3.2 | DeepSeek | $0.280 | $0.028 | $0.400 | $0.044 |
| gpt-oss-20b | Groq | $0.075 | $0.037 | $0.300 | $0.046 |
| zai-glm-5-3-flash | Mistral | $0.150 | $0.030 | $0.500 | $0.046 |
| glm-5.3-flash | Z.ai | $0.150 | $0.030 | $0.500 | $0.046 |
| deepseek-flash | DeepSeek | $0.300 | $0.0060 | $1.20 | $0.048 |
| deepseek-v4-flash | DeepSeek | $0.300 | $0.0060 | $1.20 | $0.048 |
| deepseek-v4-flash-vision-exp | DeepSeek | $0.300 | $0.0060 | $1.20 | $0.048 |
| deepseek-v4-flash | Alibaba Qwen | $0.200 | $0.040 | $0.400 | $0.054 |
| qwen-turbo | Alibaba Qwen | $0.050 | $0.050 | $0.200 | $0.054 |
| qwen-turbo-latest | Alibaba Qwen | $0.050 | $0.050 | $0.200 | $0.054 |
| gpt-5.6-luna | OpenAI | $0.200 | $0.020 | $1.20 | $0.060 |
| gpt-5.4-nano | OpenAI | $0.200 | $0.020 | $1.25 | $0.060 |
| glm-4.5-air | Z.ai | $0.200 | $0.025 | $1.10 | $0.061 |
| codestral-latest | Mistral | $0.300 | $0.030 | $0.900 | $0.061 |
| mistral-code-fim-latest | Mistral | $0.300 | $0.030 | $0.900 | $0.061 |
| mistral-code-latest | Mistral | $0.300 | $0.030 | $0.900 | $0.061 |
| MiniMax-M2 | MiniMax | $0.300 | $0.030 | $1.20 | $0.072 |
| MiniMax-M2.1 | MiniMax | $0.300 | $0.030 | $1.20 | $0.072 |
| MiniMax-M2.5 | MiniMax | $0.300 | $0.030 | $1.20 | $0.072 |
| gemini-3.1-flash-lite | $0.250 | $0.025 | $1.50 | $0.074 | |
| gpt-5-mini | OpenAI | $0.250 | $0.025 | $2.00 | $0.089 |
| gpt-oss-120b | Groq | $0.150 | $0.075 | $0.600 | $0.092 |
| gpt-4o-mini | OpenAI | $0.150 | $0.075 | $0.600 | $0.092 |
| qwen-plus | Alibaba Qwen | $0.400 | $0.052 | $1.20 | $0.093 |
| MiniMax-M3 | MiniMax | $0.300 | $0.060 | $1.20 | $0.099 |
| llama-3.1-8b | Cerebras | $0.100 | $0.100 | $0.100 | $0.100 |
| llama3.1-8b | Cerebras | $0.100 | $0.100 | $0.100 | $0.100 |
| glm-4-32b-0414-128k | Z.ai | $0.100 | $0.100 | $0.100 | $0.100 |
| gpt-oss-120b | Cerebras | $0.350 | $0.075 | $0.750 | $0.101 |
| mistral-large-3 | Mistral | $0.500 | $0.050 | $1.50 | $0.102 |
| mistral-large-latest | Mistral | $0.500 | $0.050 | $1.50 | $0.102 |
| deepseek-v3 | DeepSeek | $0.270 | $0.070 | $1.10 | $0.105 |
| devstral-latest | Mistral | $0.400 | $0.040 | $2.00 | $0.106 |
| devstral-medium-latest | Mistral | $0.400 | $0.040 | $2.00 | $0.106 |
| mistral-code-agent-latest | Mistral | $0.400 | $0.040 | $2.00 | $0.106 |
| MiniMax-M2.1-lightning | MiniMax | $0.300 | $0.030 | $2.40 | $0.108 |
| MiniMax-M2.5-lightning | MiniMax | $0.300 | $0.030 | $2.40 | $0.108 |
| gemini-2.5-flash | $0.300 | $0.030 | $2.50 | $0.109 | |
| gemini-3.5-flash-lite | $0.300 | $0.030 | $2.50 | $0.109 |
Prices verified: 2026-10-25 · 0.010 USD / 1M tokens (Meta muse-spark-1.2-contributor)
Read the numbers like this
- A plan is not a price per token. The same $20 buys 1.5B tokens of a cheap open model or 40M tokens of a frontier model, and several plans grant a different allowance per model.
- Request-based plans are converted with a stated tokens-per-request assumption, because a request count is not a token count. Treat those rows as an estimate and check your own task size.
- Credits are converted with the vendor's own formula (credits = weighted tokens / divisor, peak/off-peak multipliers applied), so the number is exact for that model but moves with time-of-day.
- Claude, ChatGPT, Cursor, Kimi and MiniMax publish no convertible allowance at all. They stay in the table, sorted last, with their price — an honest gap beats an invented number.
FAQ
Why do some rows show a token count and others say not quantified?
Because plans do not sell tokens. Some sell a dollar allowance (OpenCode, Command Code, OpenDesign), some a token volume (Synthetic, Cerebras, Baidu), some vendor credits with per-model multipliers (Z.ai, Volcengine Agent Plan), some a request count (Volcengine, Alibaba), and the biggest names sell an unquantified "usage" (Claude, ChatGPT, Cursor, Kimi, MiniMax). The Allowance column says which, so an inferred number is never mistaken for a published one.
Why not just compare the price per million tokens?
Input, cached read, cached write and output are priced separately, and their ratios differ by an order of magnitude between providers. A single blended figure at stated assumptions is the only number that compares plans honestly.
I only care about one model. What do I do?
Pick the model family first (DeepSeek, GLM, Claude, Qwen...). The page then lists every vendor that resells that family, what each plan costs, how many tokens the plan buys at your cache hit rate, and the resulting blended unit price.
How accurate is the token estimate?
Request counts are the one estimate on the page: they are multiplied by a stated tokens-per-request figure that you can adjust. Every other conversion uses a number the vendor publishes, and every row links to its source page so you can check it.
Where does this data come from?
Unit prices come from LiteLLM and OpenRouter; subscription prices and allowances are parsed from each vendor's own plan page (or declared with an anchor guard when the page only renders prices in the browser). The verified date is on every row — prices move, so re-check before committing to an annual plan.