Kimi API
Add Kimi, Moonshot AI's open-weight, multimodal models, to your app. Long-context, vision, and agentic intelligence through Puter.js — no API keys, no backend.
// npm install @heyputer/puter.js
import { puter } from '@heyputer/puter.js';
puter.ai.chat("Explain quantum computing in simple terms", {
model: "moonshotai/kimi-k3"
}).then(response => {
console.log(response.message.content);
});
<html>
<body>
<script src="https://js.puter.com/v2/"></script>
<script>
puter.ai.chat("Explain quantum computing in simple terms", {
model: "moonshotai/kimi-k3"
}).then(response => {
document.body.innerHTML = response.message.content;
});
</script>
</body>
</html>
The Kimi API for Long-Context, Agentic AI
Multimodal Vision
Kimi K2.5 and K2.6 read images alongside text, turning screenshots and UI designs into working code.
256K Context
A 256K-token context window handles entire codebases, long documents, and deep multi-turn sessions.
Agentic Coding
Kimi K2 is built for long-horizon coding and tool-driven agents, topping open-model coding benchmarks.
Low Cost
Open-weight pricing makes Kimi a cost-effective choice for high-volume and agentic workloads.
Tool & Function Calling
Native tool use, web search, and file parsing let Kimi drive multi-step agentic workflows.
Thinking Mode
Kimi K2 Thinking adds step-by-step reasoning for harder problems when you need more depth.
Kimi Models Compared
Every Kimi model available through Puter.js, with context, output limits, and per-million-token pricing. Click a model for full specs and examples.
| Model | Context | Max Output | Input / 1M | Output / 1M |
|---|---|---|---|---|
Kimi K3
moonshotai/kimi-k3 |
1M | 1M | $1.41 | $7.07 |
Kimi K2.7 Code
moonshotai/kimi-k2.7-code |
262K | 262K | $0.45 | $1.86 |
Kimi K2.7 Code Highspeed
moonshotai/kimi-k2.7-code-highspeed |
262K | 262K | $1.9 | $8 |
Kimi K2.6
moonshotai/kimi-k2.6 |
262K | 262K | $0.45 | $1.86 |
Kimi K2.5
moonshotai/kimi-k2.5 |
262K | 262K | $0.29 | $1.51 |
Kimi K2 Thinking
moonshotai/kimi-k2-thinking |
262K | 262K | $0.6 | $2.5 |
Kimi K2 0905
moonshotai/kimi-k2-0905 |
131K | 131K | $0.6 | $2.5 |
Kimi K2 Turbo Preview (legacy)
moonshotai/kimi-k2-turbo-preview |
262K | 262K | $1.15 | $8 |
Kimi K2 0711
moonshotai/kimi-k2 |
131K | 100K | $0.57 | $2.3 |
Kimi K2 0711 Preview (legacy)
moonshotai/kimi-k2-0711-preview |
131K | 131K | $0.6 | $2.5 |
Kimi K2 Instruct
moonshotai/kimi-k2-instruct |
128K | 66K | $0.7 | $2.5 |
Kimi Dev 72B (legacy)
moonshotai/kimi-dev-72b |
N/A | 131K | $0.29 | $1.15 |
Kimi V1 128K Vision Preview (legacy)
moonshotai/kimi-v1-128k-vision-preview |
131K | 131K | $2 | $5 |
Kimi V1 32K Vision Preview (legacy)
moonshotai/kimi-v1-32k-vision-preview |
33K | 33K | $1 | $3 |
Kimi V1 8K Vision Preview (legacy)
moonshotai/kimi-v1-8k-vision-preview |
8K | 8K | $0.2 | $2 |
Kimi V1 128K
moonshotai/kimi-v1-128k |
131K | 131K | $2 | $5 |
Kimi V1 32K
moonshotai/kimi-v1-32k |
33K | 33K | $1 | $3 |
Kimi V1 8K (legacy)
moonshotai/kimi-v1-8k |
8K | 8K | $0.2 | $2 |
Kimi K2.7 Code Fast
moonshotai/kimi-k2.7-code-fast |
262K | 262K | $1.86 | $7.72 |
Which Kimi Model Should You Use?
Pick the Kimi model that fits your workload — capability, latency, or cost.
Kimi K3
Kimi K3 is Moonshot AI's open-weight, multimodal model line, built for long-context, vision, and agentic work, with a 1M-token context window. Use it as your default for long-context, multimodal, and agentic workloads.
View Kimi K3 →Kimi K2.7 Code Fast
Kimi K2.7 Code Fast runs the same intelligence with up to 2.5x faster output. Reach for it in interactive, real-time, and live-coding workflows where response speed matters.
View Kimi K2.7 Code Fast →Kimi K2.7 Code
Other versions like Kimi K2.7 Code stay available for pinned, reproducible deployments and cost-sensitive workloads that don't need the very latest model.
View Kimi K2.7 Code →Use Kimi in Seconds
Add Kimi to your app with just a few lines of code.
No API keys, no backend, no configuration required.
// npm install @heyputer/puter.js
import { puter } from '@heyputer/puter.js';
puter.ai.chat("Explain quantum computing in simple terms", {
model: "moonshotai/kimi-k3"
}).then(response => {
console.log(response.message.content);
});
<html>
<body>
<script src="https://js.puter.com/v2/"></script>
<script>
puter.ai.chat("Explain quantum computing in simple terms", {
model: "moonshotai/kimi-k3"
}).then(response => {
document.body.innerHTML = response.message.content;
});
</script>
</body>
</html>
How Kimi Performs
Kimi K3 on standard evaluations, scored by Artificial Analysis.
Frequently Asked Questions
The Kimi API gives you access to Moonshot AI's open-weight, multimodal model line, built for long-context, vision, and agentic work — including Kimi K3 — through Puter.js. You can call any Kimi model from your app with no API keys and no backend.
Yes, it is free for you as the developer when you use it through Puter.js. With the User-Pays model, each of your users covers their own usage through their Puter account, so you can ship Kimi-powered apps at no cost to you.
Start with Kimi K3 as your default. Choose Kimi K2.7 Code Fast when you need lower latency for interactive workflows. Other versions like Kimi K2.7 Code suit pinned, cost-aware deployments.
Kimi K3 is priced at $1.41 per million input tokens and $7.07 per million output tokens at the API level. Through Puter.js it is free for developers under the User-Pays model.
Kimi K3 supports a context window of 1M tokens and can generate up to 1M tokens in a single response.
No. Puter.js handles authentication and infrastructure for you, so you can use the Kimi API without any API key, an Moonshot AI account, a server, or any configuration.
Yes — the Kimi API through Puter.js works with any JavaScript framework, Node.js, or plain HTML. Just include the library and start building.
Related Resources
Learn how to use Kimi and other models with Puter.js.