If you integrate LLMs, video, or image models through CometAPI, here's a heads-up on two pricing changes this week — your code is fine on both:
1️⃣ Kimi K3 — 20% off, effective immediately
Kimi K3 resources have stabilized on our platform. We're passing that on: K3 is now 80% of the official list price. Effective immediately — no action required.
👉 K3 pricing → https://bit.ly/3TXkHRK
Example call (unchanged)
curl https://api.cometapi.com/v1/chat/completions \
-H "Authorization: Bearer $COMETAPI_API_KEY" \
-d '{"model": "kimi-k3", "messages": [{"role": "user", "content": "..."}]}'
2️⃣ KLING — alignment with official list, 2026-07-29 01:00 UTC (09:00 CST)
KLING model pricing will be aligned with the official list price across all hosted scenarios.
👉 KLING pricing → https://bit.ly/4x7EhJu
What you do NOT need to change
✅ Endpoint URL ✅ Request body schema ✅ Model ID ✅ Task polling logic
If you're integrated today, you'll keep working today, tomorrow and the next day without redeploying.
Scenarios touched
- Image generation & editing
- Omni image generation
- Video generation & video effects
- Audio generation
- TTS (text-to-speech)
- Lip-sync
- Avatar video
- Virtual try-on
- Image expansion
- Recognition For most workloads (text-to-image, image-edit, TTS, recognition) the difference is minimal. Compute-heavy jobs (video generation, avatar video, virtual try-on) will see more material price adjustments — review the live page before scaling.
Example call (unchanged)
curl https://api.cometapi.com/v1/kling/video/generations \
-H "Authorization: Bearer $COMETAPI_API_KEY" \
-d '{"model": "kling-video-v2", "prompt": "..."}'
---
Bottom line: your code keeps working, your bill may shift slightly. Both updates are aligned with official pricing, so what you see on CometAPI now matches what you'd see going direct — minus the convenience of one key, one bill, one dashboard across 500+ models.
Questions? Drop a comment 👇
Top comments (0)