Home/Calculator/Context window
Prompt capacity check

Context window calculator for AI models

Check whether a prompt and planned output fit inside a model's tracked context window. Enter token counts, choose a model, and see the remaining capacity before you send the request.

Context usage
0%

Calculating...

Used0
Remaining0
Tracked limit0

View GPT-4o mini model details →

Compare tracked limits

How much context do models allow?

These limits come from the current model records. Availability, API route, output caps, and request formatting can still affect a real call.

ProviderModelContext windowChecked
OpenAIGPT-4o mini128,000 tokens2026-08-02
OpenAIGPT-4o128,000 tokens2026-08-02
OpenAIGPT-5.6 Sol1,050,000 tokens2026-08-02
OpenAIGPT-5.6 Terra1,050,000 tokens2026-08-02
OpenAIGPT-5.6 Luna1,050,000 tokens2026-08-02
OpenAIgpt-5.4-mini400,000 tokens2026-08-02
OpenAIgpt-5.4-nano400,000 tokens2026-08-02
AnthropicClaude Sonnet 51,000,000 tokens2026-08-02
AnthropicClaude Opus 4.81,000,000 tokens2026-08-02
AnthropicClaude Sonnet 4.61,000,000 tokens2026-08-02
AnthropicClaude Haiku 4.5200,000 tokens2026-08-02
GoogleGemini 2.5 Pro1,048,576 tokens2026-08-02
GoogleGemini 2.5 Flash1,048,576 tokens2026-08-02
GoogleGemini 2.5 Flash-Lite1,048,576 tokens2026-08-02
DeepSeekDeepSeek V4 Pro1,000,000 tokens2026-08-02
DeepSeekDeepSeek V4 Flash non-thinking1,000,000 tokens2026-08-02
AnthropicClaude Fable 51,000,000 tokens2026-08-02
AnthropicClaude Mythos 51,000,000 tokens2026-08-02
GoogleGemini 3.5 Flash1,048,576 tokens2026-08-02
GoogleGemini 3.5 Live Translate Preview131,072 tokens2026-08-02
GoogleGemini 3 Flash Preview1,048,576 tokens2026-08-02
GoogleGemini 3.1 Flash-Lite1,048,576 tokens2026-08-02
GoogleGemini 3.1 Pro Preview1,048,576 tokens2026-08-02
GroqGPT-OSS 20B on Groq131,072 tokens2026-07-15
GroqGPT-OSS 120B on Groq131,072 tokens2026-07-15
GroqLlama 4 Scout on Groq131,072 tokens2026-07-15
GroqQwen3-32B on Groq131,072 tokens2026-07-15
CohereCommand R 08-2024128,000 tokens2026-07-17
CohereCommand R+ 08-2024128,000 tokens2026-07-17
CohereCommand A256,000 tokens2026-07-17
MistralMistral Medium 3.5256,000 tokens2026-07-17
Moonshot AIKimi K31,048,576 tokens2026-07-17
Moonshot AIKimi K2.7 Code262,144 tokens2026-07-17
Moonshot AIKimi K2.7 Code High-Speed262,144 tokens2026-07-17
Moonshot AIKimi K2.6262,144 tokens2026-07-17
xAIGrok 4.5500,000 tokens2026-07-17
GoogleGemini 3.6 Flash1,048,576 tokens2026-08-02
GoogleGemini 3.5 Flash-Lite1,048,576 tokens2026-08-02
AnthropicClaude Opus 51,000,000 tokens2026-08-02

Use a measured count

Count a representative prompt first, then add expected output tokens. Plain word counts are not a reliable substitute for model tokenization.

Open the token counter

Check cost as well as fit

A prompt can fit several models but have very different input, output, cache, and long-context costs.

Estimate API cost

Verify the provider limit

StackLens records the checked model limit and source date. Open the model page for the source note and documented caveats.

Find a model
FAQ

Questions teams ask before choosing

What does a context window calculator show?

It compares the token count you enter with a model's tracked context-window limit and shows the remaining capacity. It does not tokenize the text itself.

Does the context window include the model output?

A model request usually shares its context budget across input and generated output, but provider behavior can vary. Treat the result as a planning check and verify the selected model's official documentation.

Why can a prompt fit in one model but not another?

Context limits differ by model and sometimes by API route or product tier. Use the model's source-tracked limit here, then check the provider documentation before sending a large request.

Related pages
tools / token countercalculator / llm api costmodels / findermodels / cost mapmethodology