PLAN THE NEXT REQUEST
Give the answer
room to arrive.
A context window has to hold more than the question. Budget the instructions, history, documents and tools, then leave space for the reply.
01 / COUNT THE WHOLE REQUEST
What needs to fit?
Enter token counts from your tokenizer or API usage. Include messages sent again on this call. The examples are illustrative workloads.
Reply allowance reserves space for the output. Extra headroom leaves space unused for message formatting or uncertain counts. It is your planning margin, not extra billable input.
02 / CHECK A RECORDED LIMIT
anthropic/claude-sonnet-4
19,000tokens to make room for
Within recorded limits
- Input
- 16,000
- Reply allowance
- 2,000
- Extra headroom
- 1,000
- Still available
- 181,000
- Planning context limit
- 200,000
- Catalogue context quote
- 1,000,000
- Top-provider context quote
- 200,000
- Top-provider output cap
- 64,000
All limits are in tokens. When the two context quotes differ, this tool uses the smaller known figure. An unknown output cap remains unknown.
OpenRouter catalogue ↗ · Checked
The same request across 39 models Compare limits ↗
Every row uses the counts above. Limits come from the same dated review. Model names are labels, so use the exact ID when configuring a request.
39 matching models.
| Model | Planning context | Output cap | Your budget |
|---|---|---|---|
| Qwen 2.5 72Bqwen/qwen-2.5-72b-instruct | 32,768 | 16,384 | Within recorded limits |
| DeepSeek R1deepseek/deepseek-r1 | 64,000 | 16,000 | Within recorded limits |
| GPT-4oopenai/gpt-4o | 128,000 | 16,384 | Within recorded limits |
| Llama 4 Maverickmeta-llama/llama-4-maverick | 128,000 | 115,200 | Within recorded limits |
| Mistral Small 3.1mistralai/mistral-small-3.1-24b-instruct | 128,000 | 102,400 | Within recorded limits |
| Kimi k2moonshotai/kimi-k2 | 131,072 | 100,352 | Within recorded limits |
| DeepSeek V3deepseek/deepseek-chat-v3-0324 | 163,840 | 147,456 | Within recorded limits |
| GLM-5z-ai/glm-5 | 198,000 | 128,000 | Within recorded limits |
| Claude Haiku 4.5anthropic/claude-haiku-4.5 | 200,000 | 64,000 | Within recorded limits |
| Claude Opus 4anthropic/claude-opus-4 | 200,000 | 32,000 | Within recorded limits |
| Claude Sonnet 4anthropic/claude-sonnet-4 | 200,000 | 64,000 | Within recorded limits |
| GLM 5.1z-ai/glm-5.1 | 200,000 | 128,000 | Within recorded limits |
| o3openai/o3 | 200,000 | 100,000 | Within recorded limits |
| o3-miniopenai/o3-mini | 200,000 | 100,000 | Within recorded limits |
| MiniMax M2minimax/minimax-m2 | 204,800 | 131,072 | Within recorded limits |
| MiniMax M2.7minimax/minimax-m2.7 | 204,800 | 131,072 | Within recorded limits |
| Kimi K2.6moonshotai/kimi-k2.6 | 262,144 | 235,929 | Within recorded limits |
| Mistral Large 3mistralai/mistral-large-2512 | 262,144 | 209,715 | Within recorded limits |
| Mistral Small 4mistralai/mistral-small-2603 | 262,144 | 209,715 | Within recorded limits |
| Nemotron 3 Supernvidia/nemotron-3-super-120b-a12b | 262,144 | 16,384 | Within recorded limits |
| Qwen3.6 27Bqwen/qwen3.6-27b | 262,144 | 65,536 | Within recorded limits |
| Qwen3.6 35B A3Bqwen/qwen3.6-35b-a3b | 262,144 | 235,929 | Within recorded limits |
| Trinity Large Thinkingarcee-ai/trinity-large-thinking | 262,144 | 80,000 | Within recorded limits |
| Llama 4 Scoutmeta-llama/llama-4-scout | 327,680 | 16,384 | Within recorded limits |
| GPT-5openai/gpt-5 | 400,000 | 128,000 | Within recorded limits |
| GPT-5 Miniopenai/gpt-5-mini | 400,000 | 128,000 | Within recorded limits |
| GPT-5 Nanoopenai/gpt-5-nano | 400,000 | 128,000 | Within recorded limits |
| Claude Fable 5anthropic/claude-fable-5 | 1,000,000 | 128,000 | Within recorded limits |
| Claude Opus 4.7anthropic/claude-opus-4.7 | 1,000,000 | 128,000 | Within recorded limits |
| Grok 4.3x-ai/grok-4.3 | 1,000,000 | 900,000 | Within recorded limits |
| GPT-4.1openai/gpt-4.1 | 1,047,576 | 32,768 | Within recorded limits |
| Gemini 2.5 Flashgoogle/gemini-2.5-flash | 1,048,576 | 65,535 | Within recorded limits |
| Gemini 2.5 Progoogle/gemini-2.5-pro-preview | 1,048,576 | 65,536 | Within recorded limits |
| Gemini 3.5 Flashgoogle/gemini-3.5-flash | 1,048,576 | 65,536 | Within recorded limits |
| GPT-5.4openai/gpt-5.4 | 1,050,000 | 128,000 | Within recorded limits |
| GPT-5.5openai/gpt-5.5 | 1,050,000 | 128,000 | Within recorded limits |
| GPT-5.5 Proopenai/gpt-5.5-pro | 1,050,000 | 128,000 | Within recorded limits |
| Gemini 2.0 Flashgoogle/gemini-2.0-flash-001 | Unknown | Unknown | Context unknown |
| Laguna XS.2poolside/laguna-xs.2 | Unknown | Unknown | Context unknown |