Token counter
Paste text to count its tokens, then push the count into the calculator below.
Token counts for non-OpenAI providers are estimates. Newer Claude/Gemini models tokenize differently (often ~30% more).
Groq cost calculator
Cost per call = input×price + output×price (cached input, where supported, is billed at the discounted cache-read rate). Monthly = cost per call × requests.
| Model | Cost / call | Cost / month |
|---|
Groq API pricing (per 1M tokens, USD)
| Model | Input | Cached input | Output |
|---|---|---|---|
| Llama 3.3 70B Versatile | $0.590 | — | $0.790 |
| Llama 3.1 8B Instant | $0.050 | — | $0.080 |
| GPT-OSS 120B | $0.150 | — | $0.600 |
| Qwen3 32B | $0.290 | — | $0.590 |
Groq hosts open-weight models at very high throughput; token prices are for the hosted API.
Verified 2026-07-11 from Groq. LLM API prices change often. These figures were verified against official pricing pages on the date above — always confirm with the provider before relying on them for budgeting.
❓ Frequently asked questions
How accurate are the token counts?
Token counts are exact for OpenAI models — we run OpenAI's o200k_base tokenizer (the same BPE used by GPT-4o/4.1/5) locally in your browser. Other providers don't publish browser-ready tokenizers, so their counts are close estimates; newer Claude and Gemini models in particular tokenize the same text into ~30% more tokens.
Is my text sent anywhere?
No. Tokenizing and cost math run entirely in your browser. Nothing you paste is uploaded or stored.
How current is the pricing?
Prices were verified on 2026-07-11 against each provider's official pricing page (linked under the table). LLM prices change often, so confirm with the provider before committing a budget.
What about cached input and batch discounts?
The calculator supports a cached-input field where the provider offers prompt-cache pricing (OpenAI, Anthropic, Gemini, DeepSeek). Batch APIs (typically ~50% off) aren't applied automatically — halve the result if you use batch.