Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Kling AI Model Catalog · jeremylongshore | DecimalAI

jeremylongshore/klingai-model-catalog

SkillScore88Maintained

Explore Kling AI models, versions, and capabilities for video and image generation. Use when selecting models or comparing features. Trigger with phrases like 'kling ai models', 'klingai capabilities', 'kling video models', 'klingai features'.

AI/MLby jeremylongshore

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

AI Engineer packsaaskling-aimodels

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/jeremylongshore-klingai-model-catalog/SKILL.md

decimalai skills pull jeremylongshore-klingai-model-catalog --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/jeremylongshore-klingai-model-catalog/SKILL.md https://app.decimal.ai/s/jeremylongshore-klingai-model-catalog/SKILL.md

Raw file: /s/jeremylongshore-klingai-model-catalog/SKILL.md ↗ · machine-readable summary at /s/jeremylongshore-klingai-model-catalog.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

25 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 5d ago

Yes — +68% lift+3% tokens0% turns88% of whole cases pass with skill (22/25)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-01	✗→✓	▲ Improved	-19%	0%
case-02	✗→✓	▲ Improved	-12%	0%
case-03	✗→✓	▲ Improved	23%	0%
case-04	✗→✓	▲ Improved	-19%	0%
case-05	✗→✓	▲ Improved	63%	0%

Showing 5 of 25 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/10/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

jeremylongshore / claude-code-plugins-plus-skills

skills/.curated/klingai-model-catalog/SKILL.md

View on GitHub ↗

License MITImported 8/25/2026Version v1Bundled 5 files

Skill contents

references/

SKILL.md

Kling AI Model Catalog

Overview

Kling AI offers multiple model versions across video generation, image generation, lip sync, virtual try-on, and effects. Each version trades off quality, speed, and cost. This skill is the reference for choosing the right model.

Video Generation Models

| Model ID | Supports | Max Duration | Resolution | Speed | Quality | |----------|----------|-------------|------------|-------|---------| | kling-v1 | T2V, I2V | 10s | 720p | Fast | Good | | kling-v1-5 | I2V only | 10s | 1080p | Fast | Better | | kling-v1-6 | T2V, I2V | 10s | 1080p | Medium | Better+ | | kling-v2-master | T2V, I2V | 10s | 1080p | Medium | High | | kling-v2-1 | I2V only | 10s | 1080p | Medium | High | | kling-v2-1-master | T2V, I2V | 10s | 1080p | Medium | High | | kling-v2-5-turbo | T2V, I2V | 10s | 1080p 30fps | Fast | High | | kling-v2-6 | T2V, I2V | 10s | 1080p 30-48fps | Medium | Highest |

T2V = text-to-video, I2V = image-to-video

Kling v2.5 Turbo (Recommended for Speed)

40% faster than v2.0
Up to 1080p at 30 FPS
Best cost/quality ratio for production pipelines

Kling v2.6 (Recommended for Quality)

Native audio generation (voice, SFX, ambient in one pass)
1080p at 30-48 FPS
Set motion_has_audio: true for synchronized audio

Image Generation Models (Kolors)

| Model ID | Purpose | Resolution | |----------|---------|------------| | kolors-v1-5 | Face/subject reference | Up to 2048x2048 | | kolors-v2-0 | Image restyle | Up to 2048x2048 | | kolors-v2-1 | Text-to-image | Up to 2048x2048 |

Specialty Models

| Feature | Endpoint | Model Versions | |---------|----------|----------------| | Lip Sync | /v1/videos/lip-sync | v1.6+ | | Virtual Try-On | /v1/images/kolors-virtual-try-on | v1.5 | | Video Extension | /v1/videos/video-extend | All video models | | Effects | /v1/videos/effects | v1.6+ | | Motion Control | T2V/I2V with camera_control | v1.6+ |

Mode Selection

Every video generation accepts a mode parameter:

| Mode | Credits (5s) | Credits (10s) | Use Case | |------|-------------|---------------|----------| | standard | 10 | 20 | Drafts, previews, iteration | | professional | 35 | 70 | Final output, client delivery |

Model Selection Decision Tree

Need fastest generation?
  → kling-v2-5-turbo + standard mode

Need highest quality?
  → kling-v2-6 + professional mode

Need audio in the video?
  → kling-v2-6 with motion_has_audio: true

Image-to-video only?
  → kling-v2-1 (optimized for I2V)

Budget-conscious production?
  → kling-v2-5-turbo + standard mode (10 credits/5s)

Legacy compatibility?
  → kling-v1-6 (stable, well-documented)

API Usage

python
# Specify model in any video generation request
response = requests.post(f"{BASE}/videos/text2video", headers=headers, json={
    "model_name": "kling-v2-6",       # model version
    "mode": "professional",            # standard or professional
    "prompt": "A futuristic city at sunset with flying cars",
    "duration": "5",
    "aspect_ratio": "16:9",
})

Aspect Ratios (All Models)

| Ratio | Use Case | |-------|----------| | 16:9 | Landscape, YouTube, presentations | | 9:16 | Vertical, TikTok, Reels, Stories | | 1:1 | Square, Instagram, thumbnails | | 4:3 | Classic TV, presentations | | 3:4 | Portrait photos | | 3:2 | Standard photography | | 2:3 | Tall portrait | | 21:9 | Ultra-wide, cinematic |

Resources

Want to see if jeremylongshore/klingai-model-catalog would activate on your phrasing? Try it before you install — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/30/2026, 18:45 UTC · 25 cases

Lift

+68%

88% vs 20%

Agent turns

1.0 vs 1.0

Tokens

+3%

2,151.76 vs 2,087.24

fail → pass · 17pass → pass · 5fail → fail · 3

Per-case results (25)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-01 I'm building an automated workflow to produce high-quality landscape videos from text prompts where realistic visual fidelity and synchronized sound effects/audio are essential. Could you supply the complete API JSON request payload along with an explanation of which model parameters and execution settings to use for optimal quality?	fail→pass	37,308	13,011	-65%	1	1	0%	3,391	2,741	-19%	0	0	—
case-02 We need to generate rapid 5-second draft previews during internal brainstorming sessions while minimizing credit usage as much as possible. Please provide a setup recommendation detailing the model, generation mode, and expected credit consumption per clip.	fail→pass	19,466	3,826	-80%	1	1	0%	2,331	2,050	-12%	0	0	—
case-03 Our team needs to implement two features: adding synchronized mouth movements to an existing video file, and generating video clips purely from image input references. Can you list the recommended API endpoints, required model choices, and input criteria for both tasks?	fail→pass	15,085	14,429	-4%	1	1	0%	2,663	3,274	+23%	0	0	—
case-04 We want to take existing user photos and perform artistic restyling (image-to-image restyle) up to 2048x2048 resolution using the Kolors model suite. A developer suggests using kolors-v1-5, but we want the specific model designated for restyling. Which model ID should be used and what resolution does it support?	fail→pass	17,221	7,744	-55%	1	1	0%	2,349	1,904	-19%	0	0	—
case-05 Our application needs to generate high-resolution still images directly from text descriptions. Some team members assumed kolors-v1-5 was appropriate, but we need the specific Kolors model version designated for text-to-image generation. Which model ID should be specified in the request?	fail→pass	11,401	7,922	-31%	1	1	0%	1,122	1,834	+63%	0	0	—
case-06 We need an image generation model from the Kolors family specifically designed for face and subject reference conditioning up to 2048x2048. Should we use kolors-v2-1 or a different dedicated version?	fail→pass	20,843	8,202	-61%	1	1	0%	2,298	1,957	-15%	0	0	—
case-07 We are building an image animation pipeline that ONLY accepts image inputs (image-to-video). We want to avoid generic text-to-video models to optimize performance and quality. Which dedicated image-to-video model from the v2 generation should we choose?	fail→pass	30,647	11,639	-62%	1	1	0%	2,520	2,325	-8%	0	0	—
case-08 Our legacy video processing pipeline requires a highly stable, well-documented version of Kling AI that supports both text-to-video and image-to-video while maintaining backward compatibility. Which model version is recommended for legacy stability?	fail→pass	16,696	9,321	-44%	1	1	0%	1,989	2,205	+11%	0	0	—
case-09 We are launching an e-commerce clothing feature where users visualize outfits on avatars (virtual try-on). Which REST API endpoint path and model version should be used for this capability?	fail→pass	19,676	7,442	-62%	1	1	0%	2,540	1,745	-31%	0	0	—
case-10 We need to extend the duration of previously generated video clips beyond their initial length. Which API endpoint path handles video extending, and which video model versions are supported for extension?	fail→pass	10,576	7,280	-31%	1	1	0%	1,885	1,705	-10%	0	0	—
case-11 Our marketing team wants to apply automated video effects to short clips via API. What is the dedicated API endpoint for video effects, and what is the minimum required model version?	fail→pass	14,688	7,055	-52%	1	1	0%	1,465	1,653	+13%	0	0	—
case-12 We want to control camera movements (such as pan, zoom, or tilt) during video generation using camera_control parameters. Which model versions support motion control?	fail→fail	18,037	8,341	-54%	1	1	0%	2,218	2,028	-9%	0	0	—
case-13 We are planning a 10-second video generation run across client accounts. We need to compare the credit cost between standard mode and professional mode for a 10-second duration clip. What are the exact credit costs for each mode?	pass→pass	17,348	7,638	-56%	1	1	0%	2,196	1,830	-17%	0	0	—
case-14 We are calculating credit budgets for 5-second video clips. How many credits does standard mode consume versus professional mode for a 5-second video?	pass→pass	14,514	2,766	-81%	1	1	0%	1,609	1,743	+8%	0	0	—
case-15 We are generating video content specifically targeted for mobile feeds on TikTok, Instagram Reels, and YouTube Shorts. Which aspect ratio string should be passed in the API request JSON?	pass→pass	9,272	7,835	-15%	1	1	0%	767	1,720	+124%	0	0	—
case-16 We are creating an ultra-wide cinematic teaser header for a desktop landing page. Which aspect ratio key should be included in the video generation payload?	pass→pass	6,865	7,417	+8%	1	1	0%	1,166	1,744	+50%	0	0	—
case-17 Our marketing app generates square video thumbnails and Instagram grid posts. Which exact aspect ratio value should be supplied to the video generation API?	pass→pass	9,669	3,068	-68%	1	1	0%	740	1,713	+131%	0	0	—
case-18 We need a speed-optimized model for high-volume production pipelines that supports up to 1080p resolution at 30 FPS and offers a 40% speed improvement over v2.0. Which specific model ID meets these speed criteria?	fail→pass	24,678	7,647	-69%	1	1	0%	4,696	1,795	-62%	0	0	—
case-19 We need maximum motion smoothness for visual effects and require framerates between 30 and 48 FPS at 1080p. Which model supports native high framerates up to 48 FPS?	fail→pass	21,508	10,354	-52%	1	1	0%	2,877	2,168	-25%	0	0	—
case-20 An engineer wants to use kling-v1-5 for text-to-video generation at 720p. Is this model suitable for text inputs, and what resolution does it produce?	fail→pass	10,008	10,593	+6%	1	1	0%	1,505	2,160	+44%	0	0	—
case-21 We need to render retro broadcast content matching classic 4:3 television displays. Which aspect ratio parameter should be configured in the API call?	fail→pass	2,259	7,228	+220%	1	1	0%	323	1,730	+436%	0	0	—
case-22 Show an example Python requests snippet sending a POST request to generate a 5-second landscape video in professional mode using the top-tier quality model. What endpoint relative path and request JSON fields should be used?	fail→pass	20,128	11,492	-43%	1	1	0%	2,998	2,608	-13%	0	0	—
case-23 When configuring asynchronous webhooks for Kling AI video generation completion, what is the exact JSON structure of the webhook HTTP POST callback sent to our server upon video completion?	fail→fail	12,419	12,172	-2%	1	1	0%	2,374	3,431	+45%	0	0	—
case-24 What specific text prompting syntax (such as weighting syntax like (masterpiece:1.2) or negative prompt keywords like 'blurry, low quality') is officially parsed by the Kling AI video generation text parser?	fail→fail	16,017	11,779	-26%	1	1	0%	2,475	3,279	+32%	0	0	—
case-25 How do we make an API call to purchase additional Kling AI pre-paid credit packages via credit card or Stripe checkout endpoint?	fail→pass	16,297	6,973	-57%	1	1	0%	1,684	2,452	+46%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 25 cases were attempted. The headline lift of +68 percentage points is the difference between those two pass rates over the 25 comparable cases.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 20%
Pass rate with the skill: 88%
Cases attempted: 25
Cases counted toward lift: 25
Cases that regressed: 0
Skill version measured: v1
Run completed: 8/30/2026, 18:45 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-01

Prompt

I'm building an automated workflow to produce high-quality landscape videos from text prompts where realistic visual fidelity and synchronized sound effects/audio are essential. Could you supply the complete API JSON request payload along with an explanation of which model parameters and execution settings to use for optimal quality?

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run88

Live evalNeeds ≥20 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns+35% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 5 days ago

Other measured skills in the registry, with their headline benchmark lift.