Context Window Budget Calculator
Paste system, retrieved docs, history and user message; see per-model fill % and how much budget is left.
·0 input tokens used
| Model | Budget used | Used | Remaining | Window |
|---|---|---|---|---|
GPT-5 openai | 0.0% of 398,976 available | 0 | 398,976 | 400,000 |
GPT-5 mini openai | 0.0% of 398,976 available | 0 | 398,976 | 400,000 |
GPT-4o openai | 0.0% of 126,976 available | 0 | 126,976 | 128,000 |
Claude Opus 4.6 anthropic | 0.0% of 198,976 available | 0 | 198,976 | 200,000 |
Claude Sonnet 4.6 anthropic | 0.0% of 198,976 available | 0 | 198,976 | 200,000 |
Claude Haiku 4.5 anthropic | 0.0% of 198,976 available | 0 | 198,976 | 200,000 |
Gemini 3 Pro google | 0.0% of 998,976 available | 0 | 998,976 | 1,000,000 |
Llama 4 405B (host) meta | 0.0% of 126,976 available | 0 | 126,976 | 128,000 |
DeepSeek V3 deepseek | 0.0% of 126,976 available | 0 | 126,976 | 128,000 |
Token counts use the OpenAI o200k tokenizer as a reference. Other providers may produce slightly different counts (see Token Counter tool). Reserved-for-output is subtracted from the window before computing budget remaining.