Model comparison
Kimi vs GLM: Kimi K3 and GLM-5.3
Choose Kimi K3 for the higher reported DeepSWE v1.1 score (67.5% vs 66.9%); GLM-5.3 has the lower fixed-token cost, $0.01840 for 10,000 input and 1,000 billed output tokens.
| Dimension | Winner | Margin | Evidence |
|---|---|---|---|
| Coding benchmark | Kimi K3 | +0.6 pp reported score gap | DeepSWE v1.1: Kimi K3 67.5% vs GLM-5.3 66.9%1 |
| Reasoning benchmark | Kimi K3 | +4.8 pp reported score gap | GPQA Diamond: Kimi K3 92.9% vs GLM-5.3 88.1%1 |
| Estimated request cost | GLM-5.3 | $0.02660 difference per example | GLM-5.3 $0.01840 vs Kimi K3 $0.0450023 |
Official benchmark results
Official benchmark results
Swipe to see all columns
| Metric | Kimi K3 | GLM-5.3 |
|---|---|---|
| DeepSWE v1.1 (Resolved)1 | 67.5% · vendor-reported, 2026 | 66.9% · vendor-reported, 2026 |
| GPQA Diamond (Pass@1)1 | 92.9% · vendor-reported, 2026 | 88.1% · vendor-reported, 2026 |
| HLE w/ tools (Pass@1)1 | 59.8% · vendor-reported, 2026 | 62.5% · vendor-reported, 2026 |
| NL2Repo-Bench1 | 58.0 · vendor-reported, 2026 | 58.0 · vendor-reported, 2026 |
Performance
Performance
Swipe to see all columns
Pricing
Pricing
Swipe to see all columns
| Metric | Kimi K3 | GLM-5.3 |
|---|---|---|
| Input / 1M tokens23 | $3.00 | $1.40 |
| Output / 1M tokens23 | $15.00 | $4.40 |
| Context window (tokens)23 | 1M | 1M |
| Max output (tokens)23 | 1M | 128K |
| API and model ID23 | moonshotai/kimi-k3 | zai-org/glm-5.3 |
Which to choose
Coding, reasoning, context and cost
Kimi K3
- Choose Kimi for coding (DeepSWE 67.5% versus 66.9%) and GPQA-oriented reasoning (92.9% versus GLM’s 88.1%); the coding lead is small.
- For long completions, Kimi’s 1M output cap exceeds GLM’s 128K; both have a listed 1M input context.
- Kimi’s $0.04500 request estimate is $0.02660 above GLM; HLE with tools also favors GLM, 62.5% versus 59.8%.
GLM-5.3
- Prefer GLM for tool-assisted HLE in the common report: 62.5% versus 59.8%; NL2Repo ties at 58.0.
- GLM’s $0.01840 fixed-token estimate is lower than Kimi’s $0.04500; its input/output rates are $1.40/$4.40.
- A 128K output cap is the main long-generation constraint compared with Kimi’s 1M, despite equal 1M input contexts.
Methodology
Methodology and sources
Kimi K3 and GLM-5.3 benchmark scores use the common maximum-effort settings in DeepSeek's official comparison table. Tool-assisted HLE is shown separately from HLE without tools. The AA rows are model-level references, while prices and token ceilings belong to the named Novita APIs; streaming speed excludes the initial wait.
Novita AI serves both Kimi and GLM and publishes this comparison from their cited public sources. Estimated charges use a fixed 10,000-input, 1,000-billed-output budget at uncached synchronous prices, including billed reasoning. No private benchmark or invoice is supplied; tax, retries, discounts and negotiated capacity are excluded.
The Kimi-versus-GLM illustration models the wait for 2,000 reasoning tokens before an answer. It combines AA's unrounded first-chunk and output-rate fields, not the rounded display values; this assumption defines a cost-independent latency example rather than a measured answer time.
- DeepSeek · V4.1 Flash Model Card huggingface.co · Checked 2026-10-15
- Kimi K3 · Novita AI novita.ai · Checked 2026-10-15
- GLM-5.3 · Novita AI novita.ai · Checked 2026-10-15
- Kimi K3 · Artificial Analysis artificialanalysis.ai · Checked 2026-10-15
- GLM-5.3 · Artificial Analysis artificialanalysis.ai · Checked 2026-10-15
Build with Novita Model APIs
Create an API key and connect your application.