AI Benchmarks & Token Estimator
Compare the models developers actually ship with, then price a workload before you write the integration. Paste a prompt to estimate its token cost, set an output length and a daily request count, and the calculator returns per-request, daily, and monthly cost for every model you tick. Your prompt never leaves this tab.
Model comparison
Context window, maximum output, and standard list price per 1M tokens for the current lineups. Sort any column, filter by provider, and add a model to the estimator below.
16 models
16 of 16 models shown.
| Speed and reasoning | |||||
|---|---|---|---|---|---|
| OpenAI GPT-6 Astra gpt-6-astra | 1.05M | 128K | $10.00 | $50.00 | |
| OpenAI GPT-6.1 Sol gpt-6.1-sol | 1.05M | 128K | $2.00 | $10.00 | |
| OpenAI GPT-6 Luna gpt-6-luna | 1.05M | 128K | $0.10 | $0.50 | |
| OpenAI GPT-4.1 gpt-4.1 | 1.05M | 32K | $2.00 | $8.00 | |
| OpenAI GPT-4o gpt-4o | 128K | 16K | $2.50 | $10.00 | |
| OpenAI GPT-4o mini gpt-4o-mini | 128K | 16K | $0.15 | $0.60 | |
| Anthropic Claude Opus 5.5 claude-opus-5-5 | 1M | 128K | $4.00 | $20.00 | |
| Anthropic Claude Sonnet 5.5 claude-sonnet-5-5 | 1M | 128K | $2.00 | $10.00 | |
| Anthropic Claude Haiku 4.5 claude-haiku-4-5 | 200K | 64K | $1.00 | $5.00 | |
| Google Gemini 3.1 Pro gemini-3.1-pro-preview | 1.05M | 64K | $2.00 Over 200K prompt tokens: $4.00 | $12.00 Over 200K prompt tokens: $18.00 | |
| Google Gemini 3 Flash gemini-3-flash-preview | 1.05M | 64K | $0.50 | $3.00 | |
| Google Gemini 2.5 Flash gemini-2.5-flash | 1.05M | 64K | $0.30 | $2.50 | |
| Open Source DeepSeek V4.1 Flash deepseek-v4-flash | 1M | 384K | $0.15 Off-peak. Peak hours cost double. | $0.60 Off-peak. Peak hours cost double. | |
| Open Source DeepSeek V4 Pro deepseek-v4-pro | 1M | 384K | $0.66 Off-peak. Peak hours cost double. | $1.98 Off-peak. Peak hours cost double. | |
| Open Source Meta Llama 3.1 405B llama-3.1-405b-instruct | 128K | Not published | Self-hosted You pay for compute, not per token. | Self-hosted | |
| Open Source Meta Llama 3.1 70B llama-3.1-70b-instruct | 128K | Not published | Self-hosted You pay for compute, not per token. | Self-hosted |
No models match those filters.
Try a different provider, or clear the search box.
- Prices are standard list rates in US dollars per 1M tokens, checked against each provider's pricing page on 5 October 2026. Batch, cached-input, and regional rates are not included.
- Only Anthropic publishes a comparative latency tier for its current lineup, so that line appears on the Claude models alone. No throughput number is estimated anywhere on this page.
- Meta Llama is open weight, so it has no list price and stays out of the estimator. Hosted copies of open weights bill at whatever rate the host sets.
Token counter and cost estimator
Paste the prompt you plan to send. The estimate uses the widely published rule of thumb of four characters per token, and nothing is uploaded.
How these numbers are produced
Every figure comes from arithmetic on the published rates above. Nothing is fetched from a pricing API, and no number on this page is a measurement taken on your machine.
- Token estimate = characters divided by 4, rounded up. This is the approximation Anthropic publishes for English text. Real tokenizers differ, and languages with denser scripts produce more tokens per character, so treat the result as a planning number.
- Cost per request = (input tokens / 1M) x input price + (output tokens / 1M) x output price. Daily cost multiplies that by your request count, and monthly cost multiplies it by 30.
- Prices change without notice. Confirm against your provider before you commit to a budget, and remember that batch requests usually cost about half as much as standard ones.