GPT Token Counter
OpenAI's GPT-6, GPT-5.6 and o-series reasoning models all share the o200k_base tokenizer, which is publicly available. TokenMath loads it in your browser, so GPT token counts here are exact, not estimates.
Universal AI Token Counter
Zero-latency, client-side BPE tokenization with real-time multi-model benchmarks.
GPT Models: Context Windows & Token Prices
| Model | Context Window | Input / 1M | Output / 1M | Token Count Accuracy |
|---|---|---|---|---|
| GPT-6 Astra | 1.05M tokens | $10 | $50 | Exact BPE (o200k_base) |
| GPT-5.6 Sol | 1.05M tokens | $4 | $20 | Exact BPE (o200k_base) |
| GPT-5.6 Terra | 1.05M tokens | $2 | $12 | Exact BPE (o200k_base) |
| GPT-5.6 Luna | 1.05M tokens | $0.2 | $1.2 | Exact BPE (o200k_base) |
| OpenAI o3 | 200k tokens | $2 | $8 | Exact BPE (o200k_base) |
| OpenAI o3-pro | 256k tokens | $20 | $80 | Exact BPE (o200k_base) |
| OpenAI o4-mini | 200k tokens | $1.1 | $4.4 | Exact BPE (o200k_base) |
| GPT-5.6 Mini | 400k tokens | $0.3 | $1.2 | Exact BPE (o200k_base) |
| OpenAI o5-mini | 200k tokens | $0.8 | $3.2 | Exact BPE (o200k_base) |
How GPT Tokenization Works
GPT models split text with the o200k_base tokenizer, a Byte-Pair Encoding (BPE) scheme with a ~200,000 tokens vocabulary. Common words map to one token, while rare words, identifiers and non-Latin scripts are split into several subword pieces. Read how BPE and SentencePiece tokenization work for the details.
- Standard English prose averages about 1.3 tokens per word (roughly 4 characters per token).
- Reasoning models (o3, o5-mini) also bill hidden reasoning tokens as output, so the visible answer understates the real output count.
- Chat requests add a few formatting tokens per message on top of the text itself.
Once you know your token count, estimate a full monthly bill with the LLM cost calculator, or see every OpenAI model price.
GPT Token Counter FAQ
How do I count GPT tokens?
Paste your prompt into the counter above with a GPT model selected. TokenMath tokenizes the text in your browser and shows the token count, context window usage and input cost. Accuracy: Exact BPE (o200k_base).
What tokenizer does GPT use?
GPT models are counted with the o200k_base tokenizer (Byte-Pair Encoding (BPE), ~200,000 tokens vocabulary). OpenAI's GPT-6, GPT-5.6 and o-series reasoning models all share the o200k_base tokenizer, which is publicly available.
How many words is 1,000 GPT tokens?
For typical English prose, 1,000 tokens is roughly 700–750 words. Code, JSON and non-English text use more tokens per word, so count the actual text when budgeting.
How much does GPT-5.6 Sol cost per token?
GPT-5.6 Sol costs $4 per 1M input tokens and $20 per 1M output tokens. That's $0.00400 per 1,000 input tokens.