GLM-5.3vsKimi K3
Across 11 shared benchmarks, GLM-5.3 leads overall: GLM-5.3 wins 6, Kimi K3 wins 5, with 0 ties and an average score difference of +4.06.
GLM-5.36 wins(55%)(45%)5 winsKimi K3
Benchmark scores
Grouped by capability, sorted by largest gap within each. 11 shared benchmarks.
Coding and Software Engineer
Kimi K3 3/5| Benchmark | GLM-5.3 | Kimi K3 | Diff |
|---|---|---|---|
| Program Bench | 195 / 5Max (With Tools) | 77.801 / 5Max (With Tools) | -58.80 |
| PostTrain Bench | 39.801 / 4Max (With Tools) | 36.603 / 4Max (With Tools) | +3.20 |
| FrontierSWE | 78.102 / 4Max (With Tools) | 81.201 / 4Max (With Tools) | -3.10 |
| DeepSWE | 66.908 / 26Max (With Tools) | 67.505 / 26Max (With Tools) | -0.60 |
| SWE-Marathon | 42.501 / 4Max (With Tools) | 422 / 4Max (With Tools) | +0.50 |
AI Agent - Tool Usage
Kimi K3 2/3| Benchmark | GLM-5.3 | Kimi K3 | Diff |
|---|---|---|---|
| Automation Bench | 48.201 / 7Max (With Tools) | 30.803 / 7Max (With Tools) | +17.40 |
| Toolathlon-Verified | 733 / 5Max (With Tools) | 76.501 / 5Max (With Tools) | -3.50 |
| Terminal-Bench 2.1 | 88.203 / 43Max (With Tools) | 88.302 / 43Max (With Tools) | -0.10 |
Agent Level Benchmark
GLM-5.3 1/1| Benchmark | GLM-5.3 | Kimi K3 | Diff |
|---|---|---|---|
| Agents' Last Exam | 28.504 / 10Max (With Tools) | 28.305 / 10Max (With Tools) | +0.20 |
General Knowledge
GLM-5.3 1/1| Benchmark | GLM-5.3 | Kimi K3 | Diff |
|---|---|---|---|
| HLE | 62.503 / 181Max (With Tools) | 5614 / 181Max (With Tools) | +6.50 |
Productivity Knowledge
GLM-5.3 1/1| Benchmark | GLM-5.3 | Kimi K3 | Diff |
|---|---|---|---|
| GDPval-AA v2 | 1,7692 / 13Max (With Tools) | 1,6866 / 13Max (With Tools) | +83 |
Specs
| Field | GLM-5.3 | Kimi K3 |
|---|---|---|
| Publisher | 智谱AI | Moonshot AI |
| Release date | 2026-08-14 | 2026-07-16 |
| Model type | Reasoning model | Reasoning model |
| Architecture | MoE | MoE |
| Parameters | 753.33B | 2.8T |
| Context length | 1M | 1M |
| Max output | 128K | 1M |
API pricing
Prices use DataLearner records when available; missing fields are not inferred.
| Item | GLM-5.3 | Kimi K3 |
|---|---|---|
| Text input | Not public | ¥20 / 1M tokens |
| Text output | Not public | ¥100 / 1M tokens |
| Cache read | Not public | ¥2 / 1M tokens |
One or both models have incomplete public pricing.
Summary
- GLM-5.3leads in:Agent Level Benchmark (1/1), General Knowledge (1/1), Productivity Knowledge (1/1)
- Kimi K3leads in:Coding and Software Engineer (3/5), AI Agent - Tool Usage (2/3)
On average across the 11 shared benchmarks, GLM-5.3 scores 4.06 higher.
Largest single-benchmark gap: GDPval-AA v2 — GLM-5.3 1,769 vs Kimi K3 1,686 (+83).
Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.