Kimi API Pricing Calculator & Complete Cost Guide
Calculate Kimi API costs for K3 (flagship), K2.6, and K2.7 Code. OpenAI-compatible API with automatic caching and web search.
Pricing TLDR
- • $5 voucher when you reach $5 cumulative recharge
- • K3 (flagship): $3.00/$15.00 per M tokens • K2.6 & K2.7 Code: $0.95/$4.00 • K2.7 Code HighSpeed: $1.90/$8.00
- • Automatic caching: $0.16-$0.38/M tokens (~80-90% savings) • Web search: $0.005/call
Kimi API Cost Calculator: Monthly Pricing
Calculate by
Input Tokens
Output Tokens
API Calls / Month
Quick Examples:
Cost Optimization:
Kimi K3 (kimi-k3)
Context
Quality
Per 1M Tokens
In: $3.00
Out: $15.00
Monthly Cost
Kimi K2.6 (kimi-k2.6)
Context
Quality
Per 1M Tokens
In: $0.95
Out: $4.00
Monthly Cost
Kimi K2.7 Code (kimi-k2.7-code)
Context
Quality
Per 1M Tokens
In: $0.95
Out: $4.00
Monthly Cost
Kimi K2.7 Code HighSpeed (kimi-k2.7-code-highspeed)
Context
Quality
Per 1M Tokens
In: $1.90
Out: $8.00
Monthly Cost
Tired of manually checking your Kimi credits?
Track your Kimi agent quotas and API credits in real-time.
Free 7-day trial. No sign-up, no credit card.
About Kimi API
What is Kimi API?
Kimi API provides access to Moonshot AI's large language models. The current flagship is Kimi K3, a reasoning model with a 1M-token context window and strong agent and coding performance. Kimi K2.6 and the coding-focused K2.7 Code (with a faster K2.7 Code HighSpeed variant) sit below it as cheaper options. The older Kimi K2.5 and Moonshot V1 series were retired on August 31, 2026. The API is fully compatible with OpenAI's SDK.
- Multimodal Reasoning: Kimi K3, K2.6, and K2.7 Code support text, image, and video input natively with thinking and non-thinking modes. Strong on agent tasks, coding, and visual understanding.
- OpenAI-Compatible API: Drop-in replacement for the OpenAI API. Use existing SDKs and tools with the api.moonshot.ai/v1 endpoint. Supports tool calls, JSON mode, and streaming.
- Automatic Context Caching: Cached input is billed at $0.16/M on K2.6, $0.19/M on K2.7 Code, and $0.30/M on K3, roughly 80-90% off cache-miss input rates. No configuration required.
When to Use Kimi API
Choose Kimi K3 for the best current quality on long-horizon agent and coding workloads. Use K2.6 for a cheaper general-purpose option, or the coding-focused K2.7 Code when you mainly need code generation. Kimi K2.5 and the Moonshot V1 series were retired on August 31, 2026, so migrate any remaining traffic to K3, K2.6, or K2.7 Code.
Ideal for
- AI coding assistants and code generation
- Autonomous AI agents with tool use
- Complex multi-step reasoning
- Cost-effective alternative to GPT-5.5 and Claude Opus 4.7
- Applications requiring built-in web search integration
Not ideal for
- Applications requiring <100ms latency
- Use cases needing context beyond 1M tokens
- Tasks requiring guaranteed deterministic outputs
Kimi API Pricing Breakdown
Getting Started
Recharge at least $1 to activate your account. When your cumulative recharge reaches $5, you receive a $5 voucher bonus, effectively doubling your first $5 investment.
- Sign up at platform.moonshot.ai
- Recharge minimum $1 to activate API access
- Receive $5 voucher at $5 cumulative recharge
- OpenAI-compatible: use existing SDKs immediately
Cost Optimization Features
Automatic Context Caching (~80-90% Savings)
Cached input is billed at $0.16/M on K2.6 up to $0.38/M on K2.7 Code HighSpeed, versus cache-miss input rates of $0.95-$3.00/M. Caching is automatic with no configuration needed. View 'context caching' costs in your console.
Web Search Integration ($0.005/call)
Enable the $web_search tool for real-time information. Charged only when search is triggered. Search results count toward token usage in subsequent calls.
Multimodal Input
K3, K2.6, and K2.7 Code accept text, image, and video input natively, with thinking and non-thinking modes. Image tokens follow the same per-million-token rates as text.
Tiered Rate Limits
Rate limits scale with cumulative recharge: Tier 1 ($10) gets 15 concurrent requests and 100 RPM. Tier 5 ($3000) gets 100 concurrent and 300 RPM.
Kimi API Monthly Cost Estimates
Light Use
$5-40/mo
• Personal projects
• <1K requests/day
• K2.6 with caching
Medium Use
$40-200/mo
• Small apps
• 1-5K requests/day
• K2.6 + K2.7 Code mix
Heavy Use
$200-800/mo
• Production apps
• 5-20K requests/day
• K2.6 with cache hits
Enterprise
$800+/mo
• Large scale
• 20K+ requests/day
• Contact sales, tier 5+
6 Kimi API Cost Optimization Tips
Use Automatic Caching
Kimi's automatic context caching drops input cost to $0.16-$0.38/M (roughly 80-90% off cache-miss rates). Design prompts with consistent system messages, retrieval context, and few-shot examples to maximize cache hits.
Choose K3, K2.6, or K2.7 Code
Pick K3 ($3.00/$15.00) when you need the best Kimi quality for long-horizon agent and coding workloads. Pick K2.6 ($0.95/$4.00) for a cheaper general-purpose model. Pick K2.7 Code ($0.95/$4.00) when your workload is mostly code generation.
Migrate Off Retired Models
Kimi K2.5 and the Moonshot V1 series (moonshot-v1-8k/32k/128k and the vision variants) were retired on August 31, 2026 and now return 404 errors. Move any remaining production traffic to K2.6, K2.7 Code, or K3.
Optimize Web Search Usage
Web search costs $0.005 per call plus token costs for search results. Only enable $web_search when real-time information is needed. Search results add to input tokens in subsequent calls.
Scale Your Rate Limits
Recharge to reach higher tiers: $10 gets 15 concurrent/100 RPM, $100 gets 50 concurrent/200 RPM. Match your tier to actual throughput needs.
Monitor with CostGoat
Track Kimi API spending per model with CostGoat. Get alerts when cache hit rates drop, when K3 usage spikes, or when web search costs exceed thresholds.
Kimi Model Selection Guide
Use Case
Code Generation & Agents
Recommended Model
Kimi K3
Latest Flagship
Monthly Cost (Est.)
~$150-700
Why This Model?
Best Kimi quality for coding, agent loops, and tool use
Use Case
Cost-Optimized Multimodal
Recommended Model
Kimi K2.6
Multimodal + Thinking
Monthly Cost (Est.)
~$60-300
Why This Model?
Cheaper multimodal alternative when K3 quality isn't needed
Use Case
Image Understanding
Recommended Model
Kimi K2.6
Native Multimodal
Monthly Cost (Est.)
~$60-300
Why This Model?
Best vision quality in the Kimi lineup
Use Case
Simple Chatbot (cheapest)
Recommended Model
Kimi K2.6
Lowest Current Cost
Monthly Cost (Est.)
~$15-70
Why This Model?
Cheapest current model for basic conversations
Use Case
Web-Augmented Tasks
Recommended Model
K2.6 + Web Search
Real-time Info
Monthly Cost (Est.)
~$70-350
Why This Model?
Current information for research tasks ($0.005/call)
Kimi API Rate Limits & Tiers
User Level
Cumulative Recharge
Concurrency
RPM
TPM
User Level
Cumulative Recharge
Concurrency
RPM
TPM
User Level
Cumulative Recharge
Concurrency
RPM
TPM
User Level
Cumulative Recharge
Concurrency
RPM
TPM
User Level
Cumulative Recharge
Concurrency
RPM
TPM
User Level
Cumulative Recharge
Concurrency
RPM
TPM
RPM: Requests Per Minute | TPM: Tokens Per Minute. Tier 0 has 1.5M tokens/day limit. Tier 1+ has unlimited daily tokens. Vouchers don't count toward cumulative recharge.
Start Tracking Your Kimi API Spending
Monitor your Kimi credit balance and K2 agent quotas. See remaining usage and reset countdowns at a glance.
Free 7-day trial. No sign-up, no credit card.
Kimi API Pricing FAQ
Common questions about Kimi API costs, billing, and optimization
