Curie
r50k_base
Structural tokenization analysis engine. Optimized for Curie inputs. Calibrate context windows and compute precise API economics in real-time.
Context
2049
Type
LLM
Encoding
r 50k base
CH44
WORD9
ENCODINGr50k_base
// HAPPYTOKENIZER: WAITING FOR INPUT STREAM
Quick Answers
What is the Curie token limit?
Curie supports a 2,049 token context window. Paste your text above to see how much of that window your prompt fills.
What tokenizer does Curie use?
Curie uses the r50k_base encoding (legacy GPT-3). This page counts tokens with the vendor's real vocabulary, so the numbers match what the API sees.
How much does Curie cost per million tokens?
Curie costs $2.00 per million input tokens and $2.00 per million output tokens, with cached input at $1.00 per million.
Model Insights
Modalities
InText
OutText
Supported Features
Streaming
Supported Endpoints
Completions
Cost Manifest
Input / 1M$2.00
Output / 1M$2.00
Cached Write$1.00
* Rates are estimated based on OpenAI standard tier. Actual billing may vary based on batch processing and enterprise agreements.
// Technical Specifications
- ARCHITECTURE ID
- curie
- TOKEN ENCODING
- r50k_base
- MAX CONTEXT
- 2,049TOKENS