Model comparison
GLM-5.2 vs Kimi K3
Kimi K3 is the stronger model overall, scoring 59.5 to 51.1 on the Noometry Index. GLM-5.2 costs 2.8× less per token, which makes it the better buy when Kimi K3's lead doesn't matter for your workload.
Last verified . 48 shared benchmarks.
Summary
- They share 48 benchmarks with published results for both. GLM-5.2 scores higher in 0 categories and Kimi K3 in 9 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Kimi K3 leads 63.0 to 42.3.
- The biggest single-benchmark swing is ProofBench: 35% for GLM-5.2 and 87% for Kimi K3.
- GLM-5.2 is cheaper at $1.40 / $4.40 per million input/output tokens, against $3 / $15 for Kimi K3.
- Kimi K3 accepts more context: 1.05M tokens versus 1M.
Side by side
| GLM-5.2 | Kimi K3 | |
|---|---|---|
| Provider | Z.ai (Zhipu) | Moonshot AI |
| Noometry Index | 51.1 | 59.5 |
| Released | 2026-06-13 | 2026-07-16 |
| Weights | Open | Open |
| Context window | 1M | 1.05M |
| Max output | 131K | 1.05M |
| Input $ / M tokens | $1.40 | $3 |
| Output $ / M tokens | $4.40 | $15 |
| Results tracked | 51 | 53 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K3 leads
GLM-5.2: 51.3 (#41), Kimi K3: 61.0 (#10)
| Benchmark | GLM-5.2 | Kimi K3 |
|---|---|---|
| DeepSWE | 43.8% | 68.5% |
| FrontierCode | 24.5% | 44.2% |
| LMArena WebDev | 1603 | 1654 |
| SciCode | 50.5% | 59.5% |
| WeirdML | 70.1% | 82.6% |
| LMArena Coding | 1485 | 1508 |
| ALE-Bench | 1,047 | 1,524 |
| SWE-bench Verified | 78.7% | — |
| FrontierSWE | — | 25.9% |
Agentic & Tool Use Kimi K3 leads
GLM-5.2: 32.4 (#63), Kimi K3: 41.8 (#20)
| Benchmark | GLM-5.2 | Kimi K3 |
|---|---|---|
| APEX-Agents | 45.2% | 50.6% |
| τ²-bench Banking | 37.1% | 37.1% |
| PostTrainBench | 31.7% | 32% |
| GBAEval | 0% | 48.3% |
| Vending-Bench 2 | 8,314 | 5,165 |
| GDP.pdf | — | 19% |
Reasoning Kimi K3 leads
GLM-5.2: 42.3 (#52), Kimi K3: 63.0 (#17)
| Benchmark | GLM-5.2 | Kimi K3 |
|---|---|---|
| ARC-AGI-2 | 22.8% | 60.4% |
| SimpleBench | 58.8% | 60.7% |
| NYT Connections (extended) | 74.3% | 93.6% |
| ARC-AGI-1 | 77% | 94.5% |
| CritPt | 20.9% | 23.4% |
| Chess Puzzles | 21% | 39% |
| LMArena Hard Prompts | 1480 | 1496 |
| Mystery Game Puzzles | 19% | 26% |
| DTBench | 93.6% | 91.2% |
| LMCA | 45.8% | 52.7% |
| Surface Evolver Bench | 55.6% | 95% |
| Epoch Capabilities Index | 151.78 | 157.45 |
| Kagi LLM Benchmark | 62.6% | — |
| EBR-Bench | 9.5% | — |
| ForecastBench | — | 61.1 |
Math Kimi K3 leads
GLM-5.2: 55.7 (#43), Kimi K3: 74.2 (#16)
| Benchmark | GLM-5.2 | Kimi K3 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 59.2% | 72.2% |
| FrontierMath Tier 4 | 29.3% | 39% |
| MathArena Final-Answer Competitions | 67.6% | 87.8% |
| OTIS Mock AIME 2024-2025 | 86.4% | 97.2% |
| ProofBench | 35% | 87% |
| LMArena Math | 1482 | 1491 |
Knowledge Kimi K3 leads
GLM-5.2: 57.1 (#40), Kimi K3: 63.2 (#21)
| Benchmark | GLM-5.2 | Kimi K3 |
|---|---|---|
| GPQA Diamond | 91.9% | 93.1% |
| SimpleQA Verified | 34.2% | 50.6% |
| LMArena Expert | 1486 | 1521 |
Multimodal Not comparable
GLM-5.2: —, Kimi K3: 37.8 (#70)
| Benchmark | GLM-5.2 | Kimi K3 |
|---|---|---|
| Blueprint-Bench 2 | — | 29.5% |
| Furniture Assembly | — | 34.2% |
Multilingual Too close to call
GLM-5.2: 55.8 (#26), Kimi K3: 56.3 (#21)
| Benchmark | GLM-5.2 | Kimi K3 |
|---|---|---|
| LMArena Non-English | 1459 | 1466 |
| LMArena Chinese | 1519 | 1529 |
| LMArena French | 1479 | 1491 |
| LMArena German | 1468 | 1488 |
| LMArena Japanese | 1451 | 1487 |
| LMArena Korean | 1445 | 1458 |
| LMArena Russian | 1466 | 1482 |
| LMArena Spanish | 1477 | 1472 |
Instruction Following Too close to call
GLM-5.2: 76.9 (#34), Kimi K3: 77.7 (#14)
| Benchmark | GLM-5.2 | Kimi K3 |
|---|---|---|
| LMArena Instruction Following | 1465 | 1483 |
Long Context Too close to call
GLM-5.2: 45.3 (#43), Kimi K3: 45.8 (#29)
| Benchmark | GLM-5.2 | Kimi K3 |
|---|---|---|
| LMArena Longer Query | 1479 | 1494 |
Writing & Preference Kimi K3 leads
GLM-5.2: 70.4 (#21), Kimi K3: 76.6 (#4)
| Benchmark | GLM-5.2 | Kimi K3 |
|---|---|---|
| LMArena Text | 1470 | 1476 |
| LMArena Creative Writing | 1462 | 1454 |
| EQ-Bench Creative Writing | 1757 | 2082 |
| EQ-Bench 4 | 1222 | 1339 |
| LMArena Multi-Turn | 1469 | 1488 |
Frequently asked questions
Is GLM-5.2 better than Kimi K3?
Kimi K3 is the stronger model overall, scoring 59.5 to 51.1 on the Noometry Index. GLM-5.2 costs 2.8× less per token, which makes it the better buy when Kimi K3's lead doesn't matter for your workload.
Which is cheaper, GLM-5.2 or Kimi K3?
GLM-5.2 is cheaper. It lists at $1.40 per million input tokens and $4.40 per million output tokens; Kimi K3 lists at $3 and $15.
Is GLM-5.2 or Kimi K3 better for coding?
Kimi K3 scores higher on coding benchmarks: 61.0 versus 51.3 in the Noometry coding category.
Which has the bigger context window?
Kimi K3 does, with 1.05M tokens against 1M.
How many benchmarks do GLM-5.2 and Kimi K3 share?
48 benchmarks have published results for both models. GLM-5.2 has 51 scored results on Noometry and Kimi K3 has 53.