GLM-5.2 vs Kimi K3
Z.ai's GLM-5.2 and Moonshot AI's Kimi K3 both target production language workloads, but they price and behave differently. Here is the side-by-side.
Pricing checked against provider documentation on . How we verify
The short answer
GLM-5.2 is the cheaper option — roughly 2.8x less on a blended workload, and it suits mit-licensed coding agents. Kimi K3 justifies its premium when you need natively multimodal long-horizon agents.
Z.ai • GLM-5
$1.40 / $4.40
/ 1M tokens (input / output)
A 744B-parameter MoE with 40B active, MIT-licensed, and tuned for long-horizon coding work. It sits between DeepSeek and Kimi on price and is the usual challenger when a cheaper model keeps looping or missing repository conventions — enough capability to finish the task without flagship pricing.
Moonshot AI • Kimi K
$3.00 / $15.00
/ 1M tokens (input / output)
At 2.8T total parameters with 104B active, Kimi K3 is the largest open-weight model in general circulation and the only one here that ingests video natively. Reasoning is always on with configurable effort. The licence is the catch: it is not a standard open licence, and organisations above $20M in revenue need a separate commercial agreement.
Input price
Kimi K3 is 2.1x the price than GLM-5.2 on input tokens.
Output price
Kimi K3 is 3.4x the price than GLM-5.2 on output tokens — usually the side that dominates the bill.
Monthly cost at three workload sizes
Standard (non-batch, non-cached) rates. Reasoning models will exceed these figures because thinking tokens bill as output.
| Workload | GLM-5.2 | Kimi K3 | Difference |
|---|---|---|---|
| Light — 1M in / 200K out | $2 | $6 | $4 |
| Moderate — 10M in / 2M out | $23 | $60 | $37 |
| Heavy — 100M in / 20M out | $228 | $600 | $372 |
Specification comparison
| Attribute | GLM-5.2 | Kimi K3 |
|---|---|---|
| Input (/ 1M tokens) | $1.40 | $3.00 |
| Output (/ 1M tokens) | $4.40 | $15.00 |
| Cached input | $0.26 | $0.30 |
| Context window | 1,000,000 tokens | 1,048,576 tokens |
| Max output | 131,072 tokens | 131,072 tokens |
| Native reasoning | Yes | Yes |
| Knowledge cutoff | 2026-02 | 2026-02 |
| Relative latency | medium | high |
| Open weights | MIT | Kimi K3 License (commercial terms above $20M revenue) |
| API model ID | glm-5.2 | kimi-k3 |
| Status | stable | stable |
GLM-5.2: Independently measured by Artificial Analysis; the vendor's own pricing page was unreachable at time of checking. Treat as approximate and confirm in the console.
Kimi K3: The custom licence requires a separate commercial agreement once the licensee and its affiliates exceed $20M revenue over any consecutive 12 months.
Choose GLM-5.2 if…
- Coding agents on large repositories
- Long-horizon refactors
- Self-hosted agent stacks
- Structured output pipelines
Choose Kimi K3 if…
- Multimodal agent workflows
- Video and image understanding at length
- Ambitious long-horizon automation
- Research requiring open weights at scale
Frequently asked
Is GLM-5.2 or Kimi K3 cheaper?
GLM-5.2 is cheaper. On a blended 3:1 input-to-output workload it costs about 2.8x less than Kimi K3.
Which has the larger context window, GLM-5.2 or Kimi K3?
Kimi K3 has the larger window at 1.05M tokens versus 1M.
Should I use GLM-5.2 or Kimi K3?
Pick GLM-5.2 for mit-licensed coding agents. Pick Kimi K3 for natively multimodal long-horizon agents. If cost dominates the decision, GLM-5.2 wins; if you need the capability ceiling, benchmark both on your own evals before committing.