Kimi K3 Is Now Available on WidelAI
Kimi K3, Moonshot AI's 2.8T-parameter open-weight flagship with a 1M-token context window, is now live on WidelAI. Here's what it does, how it's priced in credits, and how to start using it in under a minute.
WidelAI Team
Building the future of AI accessibility
Kimi K3 Is Now Available on WidelAI
Moonshot AI's new flagship is live on WidelAI today. Kimi K3 is in your model picker right now under the tag kimi-k3 — no waitlist, no separate Moonshot account, and no API keys to manage. Released on July 16, 2026, Kimi K3 is a 2.8-trillion-parameter open-weight model with a one-million-token context window and native vision, and it lands at pricing that undercuts most Western flagships. This post covers what Kimi K3 is, how it is priced in WidelAI credits, when to reach for it, and how to start using it in the next minute.
Kimi K3 is built by Moonshot AI. See the official model documentation at platform.moonshot.ai.
What Is Kimi K3?
Kimi K3 is Moonshot AI's largest and most capable model to date, and the company describes it as the largest open-weight AI system built so far. It is a sparse mixture-of-experts (MoE) design totaling 2.8 trillion parameters, with only a fraction active on any given token — the approach that lets a model this large stay efficient enough to serve at competitive prices.
The headline capabilities:
- 2.8 trillion parameters in a sparse mixture-of-experts architecture
- 1 million token context window, enough for entire codebases, large document sets, or long multi-turn sessions
- Native vision, so image understanding is built in rather than bolted on
- Open-weight release, giving the broader ecosystem the ability to study and self-host the model
- Strong results on frontier coding and reasoning benchmarks, including a reported edge over several Western flagships in front-end code generation
For teams that have been watching the open-weight space, K3 is a notable moment: frontier-level performance at a scale that was previously the exclusive domain of closed labs.
How Kimi K3 Compares
The simplest way to place Kimi K3 is against the flagships you already use on WidelAI. It sits in the same tier on capability while coming in cheaper on raw API pricing.
| Model | Provider | Context | Real API price (input / output per 1M) |
|---|---|---|---|
| Kimi K3 | Moonshot | 1M | $3.00 / $15.00 |
| Claude Fable 5 | Anthropic | 1M | $10.00 / $50.00 |
| GPT-5.6 Sol | OpenAI | 1M+ | $5.00 / $30.00 |
| Gemini 3.1 Pro | 1M | $2.00 / $12.00 |
Moonshot also offers a steep cached-input discount — about $0.30 per million cache-hit input tokens — which helps when you reuse large system prompts or documents across requests. The takeaway: K3 gives you frontier-class context and reasoning at a price that is easy to justify for high-volume work.
Kimi K3 Pricing on WidelAI
WidelAI uses a transparent credit system: 1 credit equals $0.005 of real API cost, and input and output tokens are priced separately because generating output costs far more than reading input. Your credit rate for any model is simply its real per-million price multiplied by 0.2, then expressed per 1,000 tokens:
| Model | Input (credits / 1K tokens) | Output (credits / 1K tokens) |
|---|---|---|
| Kimi K3 | 0.60 | 3.00 |
That puts Kimi K3 at the same credit rate as Claude Sonnet 5 and the Sonnet 4.x line (0.60 / 3.00) — a genuinely economical rate for a model with a million-token context window. See every model side by side on our pricing transparency page.
When to Use Kimi K3
Kimi K3 is a strong default for work that benefits from a large context window and solid reasoning without a premium price tag.
Reach for Kimi K3 when
- You are working across a large codebase or document set and want the full million-token context without paying flagship-tier rates
- The task involves front-end or general code generation, where K3 benchmarks especially well
- You want a capable vision-enabled model for mixed text-and-image inputs
- You are running higher volumes and want cost per call to stay low while keeping quality high
Consider a different model when
- You need the very deepest long-horizon agentic reasoning — Claude Fable 5 or GPT-5.6 Sol may still edge ahead on the hardest autonomous tasks
- Your workflow is already tuned to a specific provider's tool surface
Tip: Because every model on WidelAI shares one credit balance, you can route everyday work to Kimi K3 and escalate only the hardest cases to a higher-cost flagship — without juggling separate accounts or keys.
A New Provider Joins WidelAI
Kimi K3 marks the arrival of Moonshot AI as a provider on WidelAI, alongside OpenAI, Google, and Anthropic. That means more genuine choice: when you can move between GPT, Gemini, Claude, and Kimi without re-architecting anything, you pick the right model for each task on cost and quality — and you stay resilient if any single model changes price or availability. You can browse Moonshot models next to every other provider in our models library.
How to Use Kimi K3 on WidelAI
Getting started takes about a minute:
- Sign in to your WidelAI account, or create one if you are new.
- Open the chat and click the model selector.
- Choose Kimi K3 from the Moonshot group — or paste the model tag (
kimi-k3) if you are calling it programmatically. - Send your prompt.
Kimi K3 is available on the Pro plan, right next to Claude Fable 5, GPT-5.6 Sol, and Gemini 3.5 Flash. Browse the full catalog on our models page.
Why Access It Through WidelAI
You could get Kimi K3 directly from Moonshot — but then you are managing another subscription, another billing relationship, and another set of API keys, all locked to a single provider. WidelAI gives you Kimi K3 alongside the best models from OpenAI, Google, and Anthropic, under one plan and one credit balance.
That matters for more than convenience. Model choice is leverage: the ability to pick the most cost-effective model that clears your quality bar, and to switch the moment something better or cheaper ships. We wrote more about that resilience angle in Claude Fable 5 Is Now Available on WidelAI.
Get Started Today
Kimi K3 is live on WidelAI right now. Whether you need its million-token context for large-codebase work, its strength on code generation, or simply a capable flagship at an economical rate, it is one click away in your model picker.
Already a member? Open the chat, pick Kimi K3, and try it on your next task.