Kling 4.0 API: Where It Works (Tested Across 7 Hubs)

👁 121 reads

If you want to plug the newest video model directly into your code today, you might be asking about the Kling 4.0 API: where it works on launch day. The short answer is that no public developer API is serving live generations yet, but you can already control the model programmatically using official developer tooling. Here is the breakdown of what is actually running across seven major hubs.

On 15 October 2026, Kuaishou officially launched Kling 4 to everyone. Alongside the new web interface, developers naturally wanted direct endpoint access. For English-speaking teams, Kling 4.0 includes native support for English speech (complete with American, British, and Indian accents) as well as on-screen English text, placing it directly among the official list of 9 supported languages. To see what developers can actually integrate right now, I checked seven separate API platforms between 17:45 and 18:40 CET on 15 October 2026.

Kling 4.0 API: Where It Works Today

I verified the availability of Kling 4.0 across the developer platforms that typically host Kuaishou models. While several providers show listings or documentation schemas, none of the traditional pay-as-you-go API routers provide actual live inference for Kling 4.0 yet.

What Happened When I Tested fal.ai

On fal.ai, seven distinct endpoints are already published with public OpenAPI schemas: text-to-video, image-to-video, reference-to-video, keyframes-to-video, video-to-video/edit, flash/text-to-video, and flash/image-to-video. The schemas detail inputs matching the official specs, such as durations from 3 to 30 seconds, 720p and 1080p resolutions, 21:9 aspect ratios, multi-prompt per shot, up to 10 keyframes, and up to 3 voice IDs.

However, running requests against these endpoints does not trigger actual model inference yet. I sent two requests with a simple test prompt (a red cube on a table, 3 seconds). Both requests completed in 0.03 seconds of compute time, cost 0 billable units, and returned the exact same pre-made 12-second placeholder video of an astronaut at 1280×720. In short: fal.ai has published the routing structure, but it is currently serving a placeholder asset before live billing starts.

The Status of Other Third-Party Hubs

The rest of the third-party ecosystem is also waiting on upstream access:

• OpenRouter: Lists 30 video models, but the Kling lineup is strictly Kling 3.0 Pro, 3.0 Standard, and O1.
• reAPI: Both Kling 4.0 and Kling 4.0 Flash pages carry a “coming soon” label.
• kie.ai: Lists Kling 4.0 and Kling 4.0 Flash as “Coming Soon”.
• Higgsfield API: Lists Kling models up to 3.0, O3, and Omni, with no 4.0 entries.
• Official Kling Developer Portal: The developer dashboard at kling.ai/dev displays a preview card reading “Kling 4.0 API Coming Soon”. The API changelog has not been updated since 15 September 2026.

The Working Alternative: Official MCP and CLI

While an independent REST API with dedicated token billing is not active yet, programmatic automation is fully operational through the official Model Context Protocol (MCP) server and Command Line Interface (CLI). As covered in our guide to Kling 4.0 MCP and CLI for Claude and ChatGPT, these tools run using your standard account session rather than a standalone API contract.

The official CLI is distributed via npm as @klingai/cli-global (version 0.2.1, requiring Node.js 18+). It acts as a thin client over the MCP tools. It authenticates through browser-based OAuth via https://kling.ai/auth using PKCE (S256), storing tokens locally in your home directory at ~/.kling/.

You can install the tool globally and authenticate:

npm i -g @klingai/cli-global
kling login

Once authenticated, you can invoke generations using kling-video-v4_0 (3 to 30 seconds, 720p or 1080p subject to account access) or kling-video-v4_0_flash (3 to 20 seconds, 720p, first frame only):

kling text_to_video --model kling-video-v4_0 "drone shot flying over a snowy mountain peak" --duration 10 --aspectRatio 16:9 --poll 300

For agents like Claude Code, you can register the remote HTTP MCP server directly:

claude mcp add --transport http kling https://kling.ai/mcp

Crucially, running commands via MCP or CLI draws credits directly from your regular Kling AI membership plan. Kling AI shows subscription prices in USD without VAT information. For context on subscription tiers, see our breakdown of Kling 4.0 pricing and plans. I checked in the kling.ai creator on 15 October 2026: Kling 4.0 costs 15 credits per second at 1080p and 10 at 720p, so a 30-second 1080p clip is 450 credits. The Standard plan does not list Kling 4.0 in its plan benefits.

Kling 3.0 Official API Pricing for Reference

To anticipate what the Kling 4.0 developer API might cost when it officially launches, you can look at the current developer rates for Kling 3.0 on kling.ai/dev/pricing. Developer access uses units, where 1 unit equals 0.14 USD, and package bundles start at 700 USD for 5,000 units with a 180-day validity window. The official developer pricing page explicitly notes: “These plans currently don’t support Kling 4.0.” For more background on how the versions differ, review our analysis of Kling 4.0 vs Kling 3.0.

How to Check API Availability Yourself

Because third-party providers update their registries rapidly, you can inspect their endpoints directly without building an entire application:

• Inspect OpenRouter Models: Query the OpenRouter public models endpoint (https://openrouter.ai/api/v1/models) and filter the JSON payload for strings containing kling. If only Kling 3.0 models appear, 4.0 is not yet routed.
• Inspect fal.ai Schemas: You can review the public OpenAPI schema directly at https://fal.ai/models/fal-ai/kling-video/v4/text-to-video. Watch the pricing section on that page to verify if non-zero compute costs have been assigned.
• Check Official Developer Docs: Visit kling.ai/dev and monitor the Developer Changelog tab. When Kling 4.0 graduates from “Preview / Coming Soon” to general release, package purchasing terms will update there.

Frequently Asked Questions

Is there an official Kling 4.0 developer API available right now?

No, the official developer API at kling.ai/dev currently lists Kling 4.0 as coming soon. Existing API unit packages do not support generation with Kling 4.0 models yet.

Can I call Kling 4.0 via fal.ai today?

While fal.ai has published schemas for seven Kling 4.0 endpoints, requests currently return a static placeholder video of an astronaut with 0 compute units billed. Live generation is not active yet.

How can I run Kling 4.0 through code or scripts right now?

You can use the official CLI package (@klingai/cli-global) or the official MCP server. These tools authenticate with your standard Kling account via OAuth and draw from your existing account credits.

Does Kling 4.0 support English prompts and dialogue?

Yes, English is one of the 9 officially supported languages for speech and on-screen text generation. The model includes specific tuning for American, British, and Indian English accents.

Google · Your sourcesQuantslant higher in your Google - one clickAdd to preferred sources →

Sources

• Kling 4.0 Release Note
• Kling Developer Platform
• Kling Developer Pricing
• fal.ai Kling Video v4 Schema
• @klingai/cli-global on npm
• reAPI Kling 4.0 Listing
• Kling AI on YouTube: official Kling 4.0 videos (28.09-15.10.2026)

As of 15 October 2026. Kling changes plans and availability often - check kling.ai before buying.

The link to Higgsfield is a partner link: if you buy through it, QUANTSLANT earns a commission and you pay the same price. It has no influence on the content, including where the coverage is critical.