A Gemini tokenizer estimates the tokens used by text sent to Gemini models. This page uses a compatible browser encoding for local planning, not Google's provider-side tokenizer. It does not count images, audio, video, or predict thinking tokens, so Google API token-count and usage records remain the authoritative billing result.
Estimates text tokens with a compatible local encoding. Images, audio, video, thinking tokens, and Google-side tokenization are not reproduced. Use this Gemini tokenizer for text prompts and cost planning, with clear limits for images, audio, video, and thinking tokens.
Provider API usage records remain the final source of truth for billing.
Input $1.5, cached input $0.15, and output $7.5 per 1M tokens.
Checked Google official API pricing Gemini 3.6 Flash model documentation
This Gemini tokenizer uses these rates for the selected text scenario. Standard paid-tier text rates. Cache storage, Batch, Flex, Priority, grounding, media, and tool charges differ.
No. The Gemini tokenizer estimates text with a compatible local encoding and does not reproduce Google's provider-side tokenizer. It is useful for comparing prompt revisions and tracking context share without uploading content, but Google token-count and usage records remain authoritative for request limits and billing.
The Estimate label applies even when two revisions are compared on the same local basis.
No. A transcript pasted as text is not the same billing object as an audio file, and an image description is not the image itself. Use the Gemini tokenizer only for text sent as text, then consult Google documentation or API usage for image, audio, video, and thinking-token accounting.
The displayed context percentage is not a guarantee that every multimodal request shape will be accepted.
The cost estimate uses the standard paid text/image/video input rate, the text output rate including thinking tokens, and the listed text cache rate. Audio input has a different price and is not silently treated as text.
Run a no-cache and a cache-read scenario. The Gemini tokenizer cannot prove a cache hit or include cache storage charges, so the result should remain a range rather than one promised invoice amount.
Move from token shape to total workload cost, or compare another provider's accuracy boundary.
The current browser verification measured that this page renders its local tool, Estimate label, source, and exact checked date, while Google Gemini documentation defines the model and multimodal limits. The limitation is provider scope: image, audio, video, message framing, thinking tokens, and final usage were not compared with a live Google billing response and remain provider-owned evidence. Google official model documentation. Last verified: .
| Evidence | Observed result or boundary |
|---|---|
| Included measurement | Submitted text, characters, words, compatible token pieces, and text-only context share. |
| Excluded measurement | Images, audio, video, thinking tokens, provider framing, and final Gemini billing usage. |
Gemini Tokenizer questions
No. The Gemini tokenizer counts or estimates text only. Images, audio, and video require provider-specific multimodal accounting.
The local count cannot predict thinking tokens. The configured output price notes that Google's paid output rate includes them.
No. This site's Gemini tokenizer does not call the Gemini API or send prompt text to Google.
The tool uses a compatible browser encoding rather than an official Gemini billing tokenizer.
This Gemini tokenizer is suitable for local text and budget planning, but it is not for provider invoicing. Use the Gemini tokenizer for text-only scenarios. The Gemini tokenizer cannot cover media, so use Google API usage data for multimodal inputs, thinking tokens, and final billing. Provider API usage records remain the final source of truth for billing.