Kimi-K3 adalah model unggulan paling canggih dari Moonshot AI hingga saat ini, dengan 2,8 triliun parameter. Dibangun di atas mekanisme perhatian linear hibrida KDA (Kimi Delta Attention) dan Attention Residuals, model ini secara native mendukung pemahaman visual serta menyediakan jendela konteks hingga 1 juta token. Sebagai model open-source pertama di dunia pada skala 3 triliun parameter, Kimi-K3 dirancang untuk skenario kecerdasan mutakhir seperti pemrograman berjangka panjang, tugas berbasis pengetahuan, dan penalaran.
Penyedia Layanan Inferensi
Penyedia layanan inferensi untuk kimi-k3 adalah Alibaba Cloud Model Studio.
Kemampuan Model
Tiongkok (Beijing)
Kemampuan | Dukungan | Kemampuan | Dukungan |
|---|---|---|---|
Input Modality | Text Image | Output Modality | Text |
Model Experience | Supported | Function Calling | Supported |
Structured Outputs | Supported | Web Search | Unsupported |
Prefix Completion | Supported | Context Caching | Supported |
Batch Inference | Unsupported | Fine-tuning | Unsupported |
Singapura
Lingkup: Internasional
Kemampuan | Dukungan | Kemampuan | Dukungan |
|---|---|---|---|
Input Modality | Text Image | Output Modality | Text |
Model Experience | Supported | Function Calling | Supported |
Structured Outputs | Supported | Web Search | Unsupported |
Prefix Completion | Supported | Context Caching | Supported |
Batch Inference | Unsupported | Fine-tuning | Unsupported |
Jerman (Frankfurt)
Lingkup: Global
Kemampuan | Dukungan | Kemampuan | Dukungan |
|---|---|---|---|
Input Modality | Text Image | Output Modality | Text |
Model Experience | Supported | Function Calling | Supported |
Structured Outputs | Supported | Web Search | Unsupported |
Prefix Completion | Supported | Context Caching | Supported |
Batch Inference | Unsupported | Fine-tuning | Unsupported |
AS (Virginia)
Lingkup: Global
Kemampuan | Dukungan | Kemampuan | Dukungan |
|---|---|---|---|
Input Modality | Text Image | Output Modality | Text |
Model Experience | Supported | Function Calling | Supported |
Structured Outputs | Supported | Web Search | Unsupported |
Prefix Completion | Supported | Context Caching | Supported |
Batch Inference | Unsupported | Fine-tuning | Unsupported |
Lingkup: Internasional
Kemampuan | Dukungan | Kemampuan | Dukungan |
|---|---|---|---|
Input Modality | Text Image | Output Modality | Text |
Model Experience | Supported | Function Calling | Supported |
Structured Outputs | Supported | Web Search | Unsupported |
Prefix Completion | Supported | Context Caching | Supported |
Batch Inference | Unsupported | Fine-tuning | Unsupported |
Jepang (Tokyo)
Lingkup: Global
Kemampuan | Dukungan | Kemampuan | Dukungan |
|---|---|---|---|
Input Modality | Text Image | Output Modality | Text |
Model Experience | Supported | Function Calling | Supported |
Structured Outputs | Supported | Web Search | Unsupported |
Prefix Completion | Supported | Context Caching | Supported |
Batch Inference | Unsupported | Fine-tuning | Unsupported |
Hong Kong (Tiongkok)
Lingkup: Global
Kemampuan | Dukungan | Kemampuan | Dukungan |
|---|---|---|---|
Input Modality | Text Image | Output Modality | Text |
Model Experience | Supported | Function Calling | Supported |
Structured Outputs | Supported | Web Search | Unsupported |
Prefix Completion | Supported | Context Caching | Supported |
Batch Inference | Unsupported | Fine-tuning | Unsupported |
Batas Konteks
Parameter | Nilai | Parameter | Nilai |
|---|---|---|---|
Max Input Length | 1048576 | Max Output Length | 1048576 |
Context Window | 1048576 | Max Input Length (Thinking Mode) | 1048576 |
Max Output Length (Thinking Mode) | 1048576 | Max Chain-of-Thought Length | 1048576 |
Harga
Halaman ini hanya menampilkan harga asli untuk panggilan API model, tidak termasuk promosi berbatas waktu apa pun. Kunjungi Konsol Model Studio untuk penawaran promosi.
Tiongkok (Beijing)
Item Penagihan | Harga (USD) | Unit |
|---|---|---|
Input | 2,827 | Per 1 juta token |
Output | 14,133 | Per 1 juta token |
Input(Implicit Cache) | 0,283 | Per 1 juta token |
Singapura
Lingkup: Internasional
Item Penagihan | Harga (USD) | Unit |
|---|---|---|
Input | 3 | Per 1 juta token |
Output | 15 | Per 1 juta token |
Input(Implicit Cache) | 0,3 | Per 1 juta token |
Jerman (Frankfurt)
Lingkup: Global
Item Penagihan | Harga (USD) | Unit |
|---|---|---|
Input | 2,827 | Per 1 juta token |
Output | 14,133 | Per 1 juta token |
Input(Implicit Cache) | 0,283 | Per 1 juta token |
AS (Virginia)
Lingkup: Global
Item Penagihan | Harga (USD) | Unit |
|---|---|---|
Input | 2,827 | Per 1 juta token |
Output | 14,133 | Per 1 juta token |
Input(Implicit Cache) | 0,283 | Per 1 juta token |
Lingkup: Internasional
Item Penagihan | Harga (USD) | Unit |
|---|---|---|
Input | 3 | Per 1 juta token |
Output | 15 | Per 1 juta token |
Input(Implicit Cache) | 0,3 | Per 1 juta token |
Jepang (Tokyo)
Lingkup: Global
Item Penagihan | Harga (USD) | Unit |
|---|---|---|
Input | 2,827 | Per 1 juta token |
Output | 14,133 | Per 1 juta token |
Input(Implicit Cache) | 0,283 | Per 1 juta token |
Hong Kong (Tiongkok)
Lingkup: Global
Item Penagihan | Harga (USD) | Unit |
|---|---|---|
Input | 2,827 | Per 1 juta token |
Output | 14,133 | Per 1 juta token |
Input(Implicit Cache) | 0,283 | Per 1 juta token |
Batas Laju
Tiongkok (Beijing)
Parameter | Nilai |
|---|---|
RPM (Requests Per Minute) | 15.000 |
TPM (Tokens Per Minute) | 1.200.000 |
Singapura
Lingkup: Internasional
Parameter | Nilai |
|---|---|
RPM (Requests Per Minute) | 10.000 |
TPM (Tokens Per Minute) | 1.200.000 |
Jerman (Frankfurt)
Lingkup: Global
Parameter | Nilai |
|---|---|
RPM (Requests Per Minute) | 15.000 |
TPM (Tokens Per Minute) | 1.200.000 |
AS (Virginia)
Lingkup: Global
Parameter | Nilai |
|---|---|
RPM (Requests Per Minute) | 15.000 |
TPM (Tokens Per Minute) | 1.200.000 |
Lingkup: Internasional
Parameter | Nilai |
|---|---|
RPM (Requests Per Minute) | 10.000 |
TPM (Tokens Per Minute) | 1.200.000 |
Jepang (Tokyo)
Lingkup: Global
Parameter | Nilai |
|---|---|
RPM (Requests Per Minute) | 15.000 |
TPM (Tokens Per Minute) | 1.200.000 |
Hong Kong (Tiongkok)
Lingkup: Global
Parameter | Nilai |
|---|---|
RPM (Requests Per Minute) | 15.000 |
TPM (Tokens Per Minute) | 1.200.000 |