Most calculators ask how many tokens you need. This one flips the question: given a fixed monthly budget, how many tokens — and how many requests — can each model actually deliver?
100% free · no signup · browser-local
Share on XSet the money you can spend, then see how many tokens each model delivers.
Enough headroom for a 2,000-token request.
Tokens / month
215.1M
Tokens / day
7.2M
Requests @ 2K
107.5K
per month
Blended cost / 1M
$0.46
Sorted by the most tokens for the same monthly spend. Click a row to change the headline numbers.
| Model | Tokens / mo | Tokens / req | Req @ 2K | Share |
|---|---|---|---|---|
| GPT-6 Luna$0.1/1M in · $0.5/1M out | 263.2M | 4.4K | 131.6K | |
| Claude Haiku 5.5$0.1/1M in · $0.5/1M out | 263.2M | 4.4K | 131.6K | |
| Qwen 2.5 72B$0.35/1M in · $0.4/1M out | 259.7M | 4.3K | 129.9K | |
| DeepSeek Flash$0.15/1M in · $0.6/1M out | 215.1M | 3.6K | 107.5K | |
| GPT-4o Mini$0.15/1M in · $0.6/1M out | 215.1M | 3.6K | 107.5K | |
| Llama 3.3 70B (Groq/Together)$0.59/1M in · $0.79/1M out | 137M | 2.3K | 68.5K | |
| DeepSeek V4 Pro$0.66/1M in · $1.98/1M out | 63.1M | 1.1K | 31.6K | |
| Gemini 3.5 Flash-Lite$0.3/1M in · $2.5/1M out | 54.3M | 905.8 | 27.2K | |
| Grok 4.3$1.25/1M in · $2.5/1M out | 47.1M | 784.3 | 23.5K | |
| Gemini 3.8 Flash$0.75/1M in · $3.75/1M out | 35.1M | 584.8 | 17.5K | |
| GPT-5.4 Mini$0.75/1M in · $4.5/1M out | 29.6M | 493.8 | 14.8K | |
| Grok 4.7$2/1M in · $6/1M out | 20.8M | 347.2 | 10.4K | |
| Mistral Large 2$2/1M in · $6/1M out | 20.8M | 347.2 | 10.4K | |
| GPT-6.1 Sol$2/1M in · $10/1M out | 13.2M | 219.3 | 6.6K | |
| Claude Sonnet 5.5$2/1M in · $10/1M out | 13.2M | 219.3 | 6.6K | |
| GPT-4o (Omni)$2.5/1M in · $10/1M out | 12.9M | 215.1 | 6.5K | |
| Command R+ (Cohere)$2.5/1M in · $10/1M out | 12.9M | 215.1 | 6.5K | |
| Gemini 3.1 Pro (Preview)$2/1M in · $12/1M out | 11.1M | 185.2 | 5.6K | |
| Claude Opus 5.5$4/1M in · $20/1M out | 6.6M | 109.6 | 3.3K | |
| Gemini 4 Argon$4/1M in · $20/1M out | 6.6M | 109.6 | 3.3K | |
| GPT-5.5$5/1M in · $30/1M out | 4.4M | 74.1 | 2.2K | |
| GPT-6 Astra$10/1M in · $50/1M out | 2.6M | 43.9 | 1.3K | |
| Claude Fable 5.1$10/1M in · $50/1M out | 2.6M | 43.9 | 1.3K |
The blended rate weights input and output prices by your token split. It assumes every request uses the same mix and ignores prompt caching, batch discounts, and retries.
Enter the monthly amount you can actually spend. A hard ceiling is more useful than a vague target because it forces a model/volume trade-off.
Chat and RAG workloads usually send far more input tokens than they receive in output. The input-share slider adjusts the blended rate.
Cheap models buy millions of tokens but may need more retries or escalation. Frontier models buy fewer tokens with higher first-pass quality.
Take the shortlisted model into the token cost calculator and add prompt caching, batch discounts, and retry overhead before you commit.
Budget planning assumes flat list prices and a constant token mix. Production traffic is bursty; keep 15–25% headroom for retries, evals, and price changes.
Simulate daily, monthly, and annual LLM bills across every model in the catalog.
Price an agentic coding session: turns, context, tool results, cache hit rate, and output tokens.
Convert any price between per-token, per-1K, per-1M, per-word, and per-request.
We'll email you when model prices change or new tools launch. Unsubscribe anytime.