GLM-5.3 vs Kimi K3
glm-5.3kimi-k3GLM-5.3 costs $0.976972 per 1M input tokens and $3.070483 per 1M output tokens, with a 1M-token context window; Kimi K3 costs $2.093511 per 1M input tokens and $10.467555 per 1M output tokens, with a 1M-token context window.
Side by side
Prices in US dollars; limits and capabilities as published by each model’s provider.
| Spec | GLM-5.3 | Kimi K3 |
|---|---|---|
| Input/ 1M tokens | $0.976972 | $2.093511 |
| Output/ 1M tokens | $3.070483 | $10.467555 |
| Cache read/ 1M tokens | $0.181438 | $0.209351 |
| Cache write/ 1M tokens | $0 | $2.093511 |
| Discount | 30% off | 30% off |
| Contexttokens | 1M | 1M |
| Max outputtokens | 128K | 1M |
| Input types | text | text, image, video |
| Output types | text | text |
| Released | 2026-08-18 | 2026-07-16 |
| Endpoints | // | // |
Same workload
One workload of 1,000,000 input and 200,000 output tokens, priced at each model’s current rates.
- GLM-5.3
- $1.5911$0.977 input + $0.6141 output
- Kimi K3
- $4.187$2.0935 input + $2.0935 output
For this workload, GLM-5.3 costs $2.596 less than Kimi K3.
Same code
Only the model field changes.
curl https://api.cheapaimodel.ai/v1/chat/completions \
-H "Authorization: Bearer $CHEAPAIMODEL_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "glm-5.3",
"messages": [{"role": "user", "content": "Hello!"}]
}'When to choose which
Differences in the published specs, then each provider’s own description.
- GLM-5.3 has the lower input price ($0.976972 vs $2.093511 per 1M tokens).
- GLM-5.3 has the lower output price ($3.070483 vs $10.467555 per 1M tokens).
- Kimi K3 accepts image and video input; GLM-5.3 doesn’t.
- GLM-5.3 was released more recently (2026-08-18 vs 2026-07-16).
- Kimi K3 can return more output in one response (1M vs 128K tokens).
About GLM-5.3
Z.ai's flagship GLM model for complex software engineering and agentic work. It keeps GLM-5.2's base model, with post-training gains in coding, long-horizon tasks and cybersecurity, and currently accepts text input only.
About Kimi K3
Moonshot AI's flagship open-weight model, with 2.8 trillion parameters built on Kimi Delta Attention and Attention Residuals. It has native visual understanding and is designed for long-horizon coding, knowledge work and deep reasoning.