Tokens & Costs
AI API providers and their pricing
Each AI company prices its models differently. Pick a provider to see every model it offers, what each costs per million tokens, and which ones support caching, batch discounts and image input.
| Provider | Models | Input price range / 1M | Largest context | Newest model |
|---|---|---|---|---|
| OpenAI | 60 | $0.018 – $150 | 1,050,000 | GPT-6.1 Sol |
| Qwen | 53 | $0.03 – $4 | 1,048,576 | Qwen3.8 Max Prime |
| 20 | $0.05 – $2 | 1,048,576 | Gemini 3.8 Flash | |
| Mistral | 20 | $0.029 – $2 | 1,048,576 | Mistral Large 4 |
| Anthropic | 16 | $0.10 – $15 | 1,000,000 | Claude Haiku 5.5 |
| Z.ai | 16 | $0.0605 – $2.80 | 1,048,576 | GLM 5.3 Prime |
| Meta | 14 | $0.05 – $1.25 | 1,048,576 | Muse Spark 1.3 |
| DeepSeek | 11 | $0.259 – $0.9553 | 1,048,576 | DeepSeek V4.1 Flash |
| MiniMax | 8 | $0.21 – $0.55 | 1,000,192 | MiniMax M3 |
| Moonshot AI | 7 | $0.45 – $0.53 | 1,048,576 | Kimi K3 |
| xAI | 7 | $1 – $2 | 2,000,000 | Grok 4.7 |
| ByteDance Seed | 6 | $0.075 – $0.50 | 262,144 | Seed 2.1 Turbo |
| Amazon | 5 | $0.035 – $2.50 | 1,000,000 | Nova 2 Lite |
| Cohere | 5 | $0.0375 – $2.50 | 256,000 | Command A+ |
| NVIDIA | 5 | $0.0469 – $0.50 | 262,144 | Nemotron 3.5 Lightning |
| Perplexity | 5 | $1 – $3 | 200,000 | Sonar Pro Search |
Comparing specific models across providers? Use the AI model comparison table.
Prices updated 2026-10-09
About
How providers differ
Each provider here is a company that builds models and sells access to them through an API. They price differently: most offer a range from small, cheap models for simple tasks to large ones for hard reasoning. For 10 of the 16 providers here, the most expensive model costs at least ten times as much per input token as the cheapest.
Features matter as much as price. Prompt caching cuts the cost of repeated input, batch processing trades speed for a discount (8 of the 16 providers here list batch prices for at least one model), and 11 of them publish open-weight models that you can also run on your own hardware; the VRAM calculator tells you what that needs.
The list includes established providers with at least 3 priced text models in the data. Each provider page lists every model with its prices, context window and features, so you can pick within a family before comparing across them.
Further reading: Claude vs GPT vs Gemini pricing, explained with live prices, The cheapest LLM APIs right now, ranked from daily price data, What is a token in AI? A plain-English guide with real examples.
FAQ
Questions people ask
Do I need a separate API key for each provider?
Yes, if you use them directly: each provider issues its own keys and bills separately. Routing services such as OpenRouter offer one key for many providers’ models. Some providers have free tiers; the free Gemini API key guide shows Google’s.
Why do providers count tokens differently?
Each model family has its own tokenizer, so the same text becomes a different number of tokens on each one. A lower price per token doesn’t always mean a lower bill. The token counter shows the counts side by side.
How often is this page updated?
Every day, from the same price data as the rest of the site. These figures are from 2026-10-09.
Which provider is cheapest?
It depends on the model size you need. Compare like for like: small models against small models, flagship against flagship. The model comparison can filter by features and sort by price.