Moonshot AI

What is Moonshot AI?

Moonshot AI is an AI lab founded in 2023 in Beijing, China, by CEO Yang Zhilin, a co-author of XLNet. It develops the Kimi model family and Kimi assistant across web, mobile, desktop and the Moonshot AI API.

The company raised roughly $2 billion at a $20 billion valuation in May 2026, bringing total funding to about $3.77 billion. Its strategy centres on publishing model weights: K2 used a Modified MIT licence, while Kimi K3 is open-weight but not OSI open-source.

Moonshot AI at a glance

Moonshot AI main AI capabilities

• Long-context reasoning: Kimi K3 processes up to 1 million tokens for workflows requiring unusually large prompts and context.
• Agentic coding: Kimi K2.7-code targets long-horizon coding workflows that combine code generation with tool and function calling.
• General-purpose reasoning: Kimi K2.6 provides general-purpose text, image understanding and coding with a 262,144-token context window.
• Multimodal understanding: Kimi K3 accepts text, images and video, while Kimi K2.6 and K2.7-code accept text and images.
• Structured automation: All three models support tool calling, streaming and strict JSON schema structured output through an OpenAI-compatible API.

When should you choose Moonshot AI?

Choose Moonshot AI if your prompts genuinely exceed 200K tokens. Kimi K3 provides a 1 million-token context window, while Kimi K2.6 and Kimi K2.7-code support 262,144 tokens.

Choose Moonshot AI if you need agentic coding at a relatively low token price. Kimi K2.7-code costs $0.95 per 1M input tokens and $4.00 per 1M output tokens and is specialised for long-horizon coding tasks.

Choose Moonshot AI if your team wants frontier-level reasoning without paying typical US frontier-model token prices. Kimi K3 costs $3.00 per 1M input tokens and $15.00 per 1M output tokens, although reasoning tokens are also billed as output.

Choose Moonshot AI if open weights provide a useful exit path from vendor lock-in. K2 models use a Modified MIT licence, while Kimi K3 is open-weight under a bespoke licence rather than OSI open-source; for EU or US hosting of Kimi models, use alternative providers on Eden AI such as Nebius, TensorX, DeepInfra or Fireworks AI.

Moonshot AI pros and cons

Moonshot AI models, features and capabilities on Eden AI

Three Moonshot AI models are available through Eden AI: Kimi K3, Kimi K2.7-code and Kimi K2.6. All three run exclusively in the AP region, for EU or US hosting, use alternative providers on Eden AI such as Nebius, TensorX, DeepInfra or Fireworks AI.

Available Moonshot AI models

Which Moonshot AI model should you use?

• Kimi K3: Use the Kimi K3 API for 1 million-token context, advanced reasoning and video understanding, but expect higher token costs than K2.6 or K2.7-code.
• Kimi K2.7-code: Use Kimi K2.7-code for agentic and long-horizon coding at $0.95/$4.00 per 1M tokens, but it has a smaller context window and no video input.
• Kimi K2.6: Use the Kimi K2.6 API for general-purpose reasoning, coding and image understanding at $0.95/$4.00 per 1M tokens, but it lacks K3's 1M context and K2.7-code's coding specialisation.

Moonshot AI API output: what data can be generated?

The Moonshot AI API returns text-based outputs from prompts, documents, code, images, video and agent instructions.

Important note on Moonshot AI accuracy and reliability

Thinking tokens are billed as output tokens, so the real cost of a Kimi K3 request can be higher than the $3.00 per 1M input-token price suggests. Workloads with heavier reasoning can therefore generate materially more billable output.

A 1 million-token context window does not guarantee uniform recall across the entire input. Test retrieval quality, instruction retention and answer consistency on your own long documents or codebases before relying on the full Kimi K3 context window in production.

Benchmark Moonshot AI models on your real prompts and compare them against at least one other provider before committing. Eden AI lets you run side-by-side model comparisons using the same application workflow, which makes differences in quality, latency and cost easier to measure.

What can you build with Moonshot AI?

Use case 1: Long-context document and codebase analysis

Kimi K3 fits long-context analysis because its 1 million-token context window can hold very large documents, repositories or combined project materials in one request. You can send the source material, ask for targeted analysis, and return summaries, extracted fields or structured JSON. For inputs that fit comfortably inside the window, this can replace some chunking and retrieval steps. It does not remove retrieval when your corpus exceeds 1M tokens, changes frequently, or requires selective access controls and source-level filtering.

Use case 2: Agentic coding assistants and developer tools

Kimi K2.7-code fits agentic coding assistants because it is specialised for long-horizon coding tasks and supports a 262,144-token context window. Your application can provide repository context, ask the model to plan a change, let it call development tools, then return code, patches or structured results for validation. Tool and function calling handles external actions, while strict JSON schema output helps keep agent responses predictable. Automatic context caching can reduce repeated processing when the same repository context is reused across successive coding requests.

Use case 3: Cost-sensitive reasoning at scale

Kimi K2.6 fits high-volume reasoning because it costs $0.95 per 1M input tokens and $4.00 per 1M output tokens while supporting 262,144 tokens of context. You can route routine classification, analysis or generation traffic to moonshot/kimi-k2.6, evaluate the response, then escalate harder cases to moonshot/kimi-k3 when deeper reasoning or a larger context window is required. Eden AI routing and fallback can automate that provider strategy, while structured output and caching help keep downstream processing and repeated-context costs under control.

Moonshot AI use cases by industry

Why use Moonshot AI through Eden AI?

Eden AI gives you access to the Moonshot AI API through the same API, billing layer and monitoring stack used for 60+ other AI providers.

Key benefits of using Moonshot AI on Eden AI

• One integration: Use one Eden AI API key for Moonshot AI and 60+ other AI providers
• Faster model comparison: Test Moonshot AI models against alternatives on the same prompts and application data
• Provider switching: Change the model string instead of maintaining a separate integration for each provider
• Production routing: Configure routing and fallback paths when Moonshot AI is unavailable or unsuitable for a workload
• Unified cost control: Track usage, model spend and billing across Moonshot AI and your other providers in one place

One API for Moonshot AI and 60+ AI providers

You can call Kimi K3, Kimi K2.7-code and Kimi K2.6 through Eden AI using the same API and API key as your other providers. Switching providers can be as simple as changing the model string, and you do not need a separate Moonshot AI account, contract or billing relationship.

Compare Moonshot AI with other AI models

Eden AI lets you compare Moonshot AI models side by side with models from providers such as DeepSeek, Anthropic and other providers in the models catalogue. Run the same prompts against multiple models to compare output quality, context requirements, latency and token cost before selecting a production route.

Add fallback and routing for production reliability

Eden AI routing and fallback let you define alternative models when Moonshot AI cannot serve a request or does not meet a workload requirement. This matters because all Moonshot models on Eden AI run in the AP region only. A non-AP fallback provider can handle workloads requiring another region, while separate fallback rules can improve availability if a preferred model fails.

Monitor usage, billing and costs in one place

Eden AI centralises usage monitoring, cost tracking and billing across Moonshot AI and your other providers. You can compare spend by model and avoid maintaining separate provider invoices and dashboards, while Eden AI pricing gives you a common commercial layer across the models you use.

Best Moonshot AI alternatives and comparisons on Eden AI 

Moonshot AI vs DeepSeek

Moonshot AI and DeepSeek both offer open-weight models with strong reasoning capabilities. Kimi K3 and DeepSeek V4 Pro/Flash each support a 1M-token context window, but DeepSeek is cheaper: V4 Pro costs $1.32/$3.96 per 1M input/output tokens and V4 Flash $0.44/$1.32, versus Kimi K3 at $3/$15. 

Moonshot wins when you need K3’s native image/video input or configurable reasoning effort; DeepSeek is the stronger option when minimising token cost for long-context text reasoning is the priority.

Moonshot AI vs Qwen

Moonshot AI offers three Kimi models on Eden AI, while Qwen offers 97 models, giving you much broader choice across model sizes, specialisations and modalities. Both include models with 1M-token context windows: Kimi K3 costs $3/$15 per 1M input/output tokens, while Qwen options shown on Eden AI start as low as $0.03/$0.26 and include multimodal models such as Qwen3.8-Max. 

Moonshot is stronger when you specifically want the Kimi architecture and K3 reasoning; Qwen wins when model breadth, EU availability or lower-cost routing matters more.

Moonshot AI vs Anthropic

Moonshot AI and Anthropic both offer models with 1M-token context windows, but the trade-off is mainly price versus deployment flexibility. Kimi K3 costs $3/$15 per 1M input/output tokens, compared with $5/$25 for Claude Opus 5 and $2/$10 for Claude Sonnet 5. 

Anthropic wins when your team needs GLOBAL availability, broader enterprise procurement acceptance or Claude-specific behaviour. Moonshot is more attractive when you want Kimi K3’s open-weight architecture, video input and lower pricing than Opus-class models.

Moonshot AI vs Lilac

Moonshot AI is the model maker and provides first-party access on Eden AI to the full available Kimi range: Kimi K3, Kimi K2.7-code and Kimi K2.6. Lilac is a third-party hosting option for Kimi K2.6 using warm GPU capacity, so Kimi K2.6 is reachable through both providers on Eden AI. Choose Moonshot when you want the broader Kimi family or K3 specifically; consider Lilac when you only need K2.6 and its hosting characteristics better match your workload. They should not be treated as interchangeable providers.

Frequently asked questions about Moonshot AI on Eden AI