Moonshot AI's upgraded Kimi flagship: a million-token context, image input and built-in reasoning for big codebases, long reports and financial models, with no Moonshot account.
Included with every account and refills every 5 hours. No card required.
You could go straight to Moonshot AI. Then you'd need a second subscription for Kimi K3, a third for GPT-6 Astra, and a fourth for image and video generation.
Most real projects use more than one model — you draft on something fast and cheap, then finish on something expensive. Plurel puts eight video models and four image models on a single balance, so you switch models instead of switching accounts.
Kimi K3.1 Pro is Moonshot AI's upgraded flagship chat model, released in October 2026 as a post-trained improvement on Kimi K3. It is built for long, careful work: large codebases, front-end pages and 3D scenes, financial models and research. It reads text and images in a context window of about 1 million tokens, reasons before it answers, and can return structured JSON. Moonshot reports that it makes far fewer unsupported claims than Kimi K3, with a claim-level hallucination rate of 8.3% against 23.4% on its own hard benchmark. On Plurel you can chat with Kimi K3.1 Pro instantly with no API key or separate subscription.
Moonshot reports 61.2% on its long-horizon coding benchmark, up from 55.95% for Kimi K3, for work that spans a whole repository.
Writes the code for 3D scenes, CAD-style parametric models and interactive front-end pages, not just snippets.
Moonshot reports gains of 7.2 points on financial modelling and 11.6 on financial research over Kimi K3.
Moonshot reports a claim-level hallucination rate of 8.3% on its hard benchmark, against 23.4% for Kimi K3.
Keep a large codebase, a long contract or a research pile in view for the whole conversation.
Roughly 750,000 words of input: long documents, long conversations, whole projects.
It thinks through a problem before it answers, which suits code, maths and multi-step analysis.
Moonshot's model takes images alongside text in the same message.
Asks for JSON and returns clean, parseable results.
Billed by what you use: about 0.35 credits for a typical message.
A typical Kimi K3.1 Pro message is about 0.35 credits — billed by length.
$7.99
100 credits
Around 285 messages — more if they are short, fewer if you paste long documents.
$14.99/mo
300 credits every month
Around 850 messages a month. Credits roll over.
Chat is included with every account and refills every 5 hours.
Open Chat
In the Plurel app, switch to the Chat tab.
Pick Kimi K3.1 Pro
Select it from the model row at the top of the chat.
Ask for real work
Paste code, a long document or a table, and say what you need.
Keep refining
The whole conversation stays in context, so follow up as you go.
Here is a diff: [paste]. List bugs, risky changes and missing tests in order of severity, then suggest the smallest fix for each.
Build a single-file HTML landing page for a bike repair shop: sticky header, hero with a booking button, three service cards, a pricing table and a footer. Clean modern CSS, no frameworks.
Write a self-contained three.js scene of a rotating low-poly planet with a small orbiting moon, soft lighting and orbit controls, as one HTML file I can open in a browser.
Build a 3-year monthly cash-flow model for a coffee-subscription start-up: assumptions table, revenue, cost of goods, payroll, runway and the break-even month. Give it as a table I can paste into a spreadsheet.
Here is a 60-page contract: [paste]. For each party, summarise obligations, deadlines, termination terms and anything unusual, in a table with page references.
Here is a claim and a source article: [paste both]. Tell me which statements the source actually supports, which it does not, and which cannot be judged from the text.
No Moonshot account needed. Free to sign up, no card.
Chat with Kimi K3.1 Pro