Last updated: October 20, 2026 · Prices checked against Moonshot AI's list prices (via the LiteLLM price file), cloud provider model cards and Oxyy's live Kimi pricing page on this date.
Disclosure: This guide is published by the Oxyy.ai team. Oxyy is an independent gateway, not affiliated with Moonshot AI.
Quick answer: On Moonshot AI's own API, Kimi K3 costs $3.00 per 1M input tokens and $15.00 per 1M output tokens, while Kimi K2.7 Code and Kimi K2.6 cost $0.95 / $4.00. Through Oxyy.ai, Kimi K2.7 Code and K2.6 currently cost $0.38 / $1.60, which is 60% below Moonshot's price and cheaper than all 16 other K2.7 Code hosts we checked.
| Kimi model | Moonshot official (input / output per 1M) | Oxyy.ai 60% off | Cached input | Context |
|---|---|---|---|---|
Kimi K3 (kimi-k3) | $3.00 / $15.00 | Not on Oxyy yet | $0.30 | 1M |
Kimi K2.7 Code (kimi-k2.7-code) | $0.95 / $4.00 | $0.38 / $1.60 | $0.19 → $0.076 on Oxyy | 262K |
Kimi K2.6 (kimi-k2.6) | $0.95 / $4.00 | $0.38 / $1.60 | $0.16 | 262K |
Moonshot list prices from the LiteLLM price file (updated October 19–20, 2026). Oxyy prices from oxyy.ai/moonshot, checked October 20, 2026; the 60% discount is a current launch promotion.
Key Takeaways
- Kimi K3 is Moonshot's 2.8-trillion-parameter flagship: $3.00 / $15.00 per 1M tokens, 1M context, always-on reasoning.
- Kimi K2.7 Code is the coding-focused model: $0.95 / $4.00 per 1M tokens, 262K context.
- Kimi K2.6 is the multimodal long-horizon model at the same $0.95 / $4.00 price.
- Oxyy charges $0.38 / $1.60 for both K2.7 Code and K2.6, 44% below the cheapest other K2.7 Code host we found (DeepInfra, $0.68 / $3.40).
- Caching saves a lot: Kimi caches repeated prefixes automatically; cached input on K2.7 Code costs $0.076 per 1M on Oxyy.
- The Moonshot API is pay-as-you-go with no subscription. It's billed separately from the Kimi membership and Kimi Code products.
How Moonshot AI (Kimi) API Pricing Works
The Kimi API platform bills input and output tokens separately, pay-as-you-go, with no subscription plan. A few rules matter for your bill:
- Automatic context caching. Kimi caches repeated request prefixes automatically. Input that hits the cache is billed only at the cached-input price.
- Cache writes on K3. For the K3 series, cache writes are billed separately by time-to-live: a 5-minute tier (the default) and a 1-hour tier. Each cache hit refreshes the entry's lifetime with no extra write charge.
- File APIs are temporarily free. Uploading and extracting a document costs nothing, but the extracted text you send to the model is billed as input.
- Taxes are extra. Listed prices exclude applicable taxes.
- Tokens: for typical English text, one token is roughly 3–4 characters.
Kimi K3 API Pricing
Kimi K3 is Moonshot's flagship: a 2.8-trillion-parameter open-weight model built on Kimi Delta Attention, with native visual understanding, always-on reasoning and a 1,048,576-token context window. It's aimed at long-horizon coding, knowledge work and deep reasoning.
Kimi K3 Price by Provider (per 1M tokens)
| Provider | Input | Output | Cache read |
|---|---|---|---|
| Sail | $2.50 | $12.50 | $0.25 |
| SCX.ai | $2.83 | $14.13 | $0.28 |
| DeepInfra | $2.85 | $14.25 | $0.285 |
| Moonshot AI (official) | $3.00 | $15.00 | $0.30 |
| Together, Fireworks, Baseten, Databricks, Novita, Cloudflare | $3.00 | $15.00 | $0.30 |
| Amazon Bedrock (Global CRIS) | $3.00 | $15.00 | $0.30 |
| Amazon Bedrock (US / India CRIS) | $3.30 | $16.50 | $0.33 |
Sources: LiteLLM price file (October 17, 2026) and the Amazon Bedrock Kimi K3 model card. Bedrock also offers a Flex tier at 0.5× and a Priority tier at 1.75× the Standard rate.
Worth knowing: at $3.00 / $15.00, Kimi K3 costs more per token than Claude Opus 5.5 does on Oxyy ($1.60 / $8.00) and about the same as Claude Sonnet 4.6 at Anthropic's list price. If you're choosing K3 for cost, benchmark it against those first.
Kimi K2.7 Code API Pricing
Kimi K2.7 Code is Moonshot's coding-focused model, built to complete end-to-end programming tasks over long contexts. It has a 262,144-token context window and supports function calling, structured output, image and video input, reasoning, prompt caching and web search.
Kimi K2.7 Code Price by Provider (per 1M tokens)
| Provider | Input | Output | Cache read |
|---|---|---|---|
| Oxyy.ai | $0.38 | $1.60 | $0.076 |
| DeepInfra | $0.68 | $3.40 | $0.136 |
| Weights & Biases | $0.71 | $3.50 | $0.15 |
| Inference APIs | $0.89 | $4.42 | $0.18 |
| Moonshot AI (official) | $0.95 | $4.00 | $0.19 |
| Alibaba Model Studio, Azure AI Foundry, Cloudflare, Fireworks, Together, Novita, Baseten | $0.95 | $4.00 | $0.16–$0.19 |
Kimi K2.6 API Pricing
Kimi K2.6 is Moonshot's multimodal model for long-horizon coding, coding-driven UI/UX generation and multi-agent orchestration, with a 262K context window.
| Provider | Input | Output | Cache read |
|---|---|---|---|
| Oxyy.ai | $0.38 | $1.60 | — |
| Weights & Biases | $0.65 | $3.41 | $0.15 |
| DeepInfra | $0.75 | $3.50 | $0.15 |
| Novita AI | $0.80 | $3.40 | $0.16 |
| Moonshot AI (official) | $0.95 | $4.00 | $0.16 |
| Together AI | $1.20 | $4.50 | $0.20 |
What Will Kimi Cost You? (Real Examples)
1,000 requests (2,000 input + 500 output tokens each)
| Model / provider | Cost |
|---|---|
| Kimi K3 (Moonshot) | $13.50 |
| Kimi K2.7 Code (Moonshot) | $3.90 |
| Kimi K2.7 Code (DeepInfra) | $3.06 |
| Kimi K2.7 Code (Oxyy) | $1.56 |
Monthly coding workload (100M input + 10M output tokens)
| Model / provider | Monthly cost | Yearly cost |
|---|---|---|
| Kimi K3 (Moonshot) | $450.00 | $5,400 |
| Kimi K2.7 Code (Moonshot) | $135.00 | $1,620 |
| Kimi K2.7 Code (DeepInfra) | $102.00 | $1,224 |
| Kimi K2.7 Code (Oxyy) | $54.00 | $648 |
Switching K2.7 Code from Moonshot direct to Oxyy saves $972 a year on this workload, and that's before prompt caching.
How to Call Kimi K2.7 Code Through Oxyy
.env
OXYY_API_KEY=sk-oxyy-your-keyPython (OpenAI SDK)
from openai import OpenAI
import os
client = OpenAI(base_url="https://api.oxyy.ai/v1", api_key=os.environ["OXYY_API_KEY"])
resp = client.chat.completions.create(
model="kimi-k2.7-code",
messages=[{"role": "user", "content": "Write unit tests for this module."}],
)
print(resp.choices[0].message.content)cURL
curl https://api.oxyy.ai/v1/chat/completions \
-H "Authorization: Bearer $OXYY_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model": "kimi-k2.6", "messages": [{"role": "user", "content": "Hello Kimi"}]}'In Cline, Kilo Code or other coding agents: choose "OpenAI Compatible", set the base URL to https://api.oxyy.ai/v1, paste your Oxyy key and use kimi-k2.7-code as the model ID.
Kimi API vs Kimi Membership vs ClinePass
| Option | How you pay | Best for |
|---|---|---|
| Kimi API (Moonshot platform) | Per token, pay-as-you-go | Apps and agents you build yourself |
| Kimi membership / Kimi Code | Subscription (separate product) | Using Kimi's own apps |
| ClinePass ($9.99/month) | Flat fee with usage windows on selected open models, including Kimi | Cline users who code most days |
| Oxyy | Per token, 60% off K2.7 Code and K2.6 | Lowest per-token price in any tool, one key for 170+ models |
Why Use Oxyy for Kimi?
- $0.38 / $1.60 per 1M tokens on Kimi K2.7 Code and K2.6, the lowest price we found across 16+ providers
- Cached input at $0.076 per 1M on K2.7 Code
- One key for Kimi plus GLM, DeepSeek, Claude, GPT, Gemini and Grok, so you can A/B test models by changing one string
- No subscription, no minimum, credits never expire, and a free starting balance with no card required
- Prices checked daily on oxyy.ai/moonshot
- About 50 billion tokens and 5.8 million requests processed across Oxyy in the last 30 days
Final Verdict
For coding work, Kimi K2.7 Code is the value pick in Moonshot's lineup at $0.95 / $4.00, and Oxyy brings it down to $0.38 / $1.60. Kimi K3 is Moonshot's most capable model, but at $3.00 / $15.00 it costs more than Claude Opus 5.5 on Oxyy, so compare carefully before committing.
Try Kimi K2.7 Code for $0.38 / $1.60 with a free Oxyy balance →
Related: Kimi K2.7 Code on Oxyy · Kimi K2.6 · GLM API pricing · DeepSeek API pricing
Frequently Asked Questions
How much does the Kimi K2 API cost?
Kimi K2.7 Code and Kimi K2.6 both cost $0.95 per 1M input tokens and $4.00 per 1M output tokens on Moonshot AI. On Oxyy they currently cost $0.38 / $1.60.
How much does the Kimi K3 API cost?
Kimi K3 costs $3.00 per 1M input tokens and $15.00 per 1M output tokens on Moonshot AI, with cache reads at $0.30 per 1M. The cheapest provider we found was Sail at $2.50 / $12.50.
What is the cheapest Kimi K2.7 Code API?
Oxyy.ai, at $0.38 input and $1.60 output per 1M tokens during its current 60% promotion. The next cheapest host we found was DeepInfra at $0.68 / $3.40.
What is the context window of Kimi K2.7 Code?
Kimi K2.7 Code has a 262,144-token context window. Kimi K3 has a 1,048,576-token context window.
Does the Kimi API support prompt caching?
Yes. The Kimi API caches repeated prefixes automatically and bills cache hits at the cached-input price. Cached input on Kimi K2.7 Code costs $0.19 per 1M on Moonshot and $0.076 on Oxyy.
Is the Kimi API subscription-based?
No. Moonshot's Kimi API is pay-as-you-go with no subscription. The Kimi membership and Kimi Code are separate products.
Is Kimi K3 available on Oxyy?
Not yet. Oxyy currently offers Kimi K2.7 Code and Kimi K2.6. Kimi K3 is available directly from Moonshot and on cloud providers such as Amazon Bedrock, Together and Fireworks.
Can I use Kimi with the OpenAI SDK?
Yes. Through Oxyy, set the base URL to https://api.oxyy.ai/v1, use your Oxyy API key and pass kimi-k2.7-code or kimi-k2.6 as the model.
Kimi K2.7 Code vs Kimi K2.6: which should I use?
Use Kimi K2.7 Code for end-to-end programming tasks and coding agents. Use Kimi K2.6 for multimodal work such as UI generation from images and multi-agent orchestration. Both cost the same.
Are Kimi API prices including tax?
No. Moonshot's listed prices exclude applicable taxes, which are calculated at checkout based on your jurisdiction.