LLM Token Counter
Optional provider counting sends text to the provider
Why: Local estimates stay in your browser. Provider counting calls the selected provider directly only when requested.
Data sent:
- Input text, selected model identifier and your provider API key
Data handling: OpenAI, Anthropic or Google receives the counting request under its own policies. The app server does not relay it; the provider key stays in page memory.
This app does not log, trace or save request/response payloads or uploaded files on its server. External APIs and storage providers apply their own retention policies; this app cannot delete their copies.
Audit the source: src/lib/provider-token-count.ts
GPT-5.6 Sol
Rough heuristic
Rough character-based heuristic with no measured accuracy guarantee. Language, code and model vocabulary can change the count substantially. Use the provider count endpoint for a request-specific count.
Text
Optional provider count
Sends one user message directly to OpenAI. Your key stays in page memory, is sent only to that provider, and is excluded from saved/shared state. Provider terms and CORS limits apply.
Context budget
Input text
0 tokens
+ System prompt
+ Conversation history
0 / 1,050,000
GPT-5.6 Sol (OpenAI) lists a 1,050,000-token context window and 128,000-token maximum output. At $4.00 per million input tokens, this input budget costs about $0.0000. Output and request overhead are additional; verify provider limits and pricing. Standard text ≤272K input. Promotional $4/$20 at least through 2026-11-21; longer prompts cost $8/$30. Cache writes cost 1.25× input. Output includes reasoning tokens.
Standards, sources and review dates
- OpenAI models and prices · selected October 2026 records · Reviewed
Only rows with individual review dates have refreshed metadata. Historical rows are unverified. Local BPE/heuristics are not API billing counts. Review interval: 7 days.
- Claude models and prices · 5.5 family + Fable 5.1 · Reviewed
First-party standard text tier; long-context, caching, residency and batch rates differ. Review interval: 7 days.
- Gemini models and prices · 4 Argon / 3.8 Flash · Reviewed
Dated promotional rates and expiry shown per row; multimodal and storage charges differ. Review interval: 7 days.