Model comparison
GPT-5.2 vs Kimi K3
Kimi K3 is the stronger model overall, scoring 59.5 to 54.1 on the Noometry Index.
Last verified . 41 shared benchmarks.
Summary
- They share 41 benchmarks with published results for both. GPT-5.2 scores higher in 1 category and Kimi K3 in 9 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in math, where Kimi K3 leads 74.2 to 60.0.
- The biggest single-benchmark swing is ProofBench: 15% for GPT-5.2 and 87% for Kimi K3.
- GPT-5.2 is cheaper at $1.75 / $14 per million input/output tokens, against $3 / $15 for Kimi K3.
- Kimi K3 accepts more context: 1.05M tokens versus 400K.
- Kimi K3 has downloadable open weights; the other is API-only.
Side by side
| GPT-5.2 | Kimi K3 | |
|---|---|---|
| Provider | OpenAI | Moonshot AI |
| Noometry Index | 54.1 | 59.5 |
| Released | 2025-12-11 | 2026-07-16 |
| Weights | Proprietary | Open |
| Context window | 400K | 1.05M |
| Max output | 128K | 1.05M |
| Input $ / M tokens | $1.75 | $3 |
| Output $ / M tokens | $14 | $15 |
| Results tracked | 67 | 53 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K3 leads
GPT-5.2: 51.6 (#37), Kimi K3: 61.0 (#10)
| Benchmark | GPT-5.2 | Kimi K3 |
|---|---|---|
| LMArena WebDev | 1416 | 1654 |
| WeirdML | 72.2% | 82.6% |
| LMArena Coding | 1447 | 1508 |
| ALE-Bench | 1,294 | 1,524 |
| SWE-bench Verified | 73.8% | — |
| DeepSWE | — | 68.5% |
| FrontierCode | — | 44.2% |
| SWE-bench Verified (bash only) | 72.8% | — |
| SWE-bench Multilingual | 66.7% | — |
| FrontierSWE | — | 25.9% |
| SciCode | — | 59.5% |
| GSO | 27.4% | — |
| AlgoTune | 2.05 | — |
Agentic & Tool Use Kimi K3 leads
GPT-5.2: 40.2 (#24), Kimi K3: 41.8 (#20)
| Benchmark | GPT-5.2 | Kimi K3 |
|---|---|---|
| τ²-bench Banking | 32.2% | 37.1% |
| Vending-Bench 2 | 3,591 | 5,165 |
| Terminal-Bench | 64.9% | — |
| APEX-Agents | — | 50.6% |
| Berkeley Function Calling Leaderboard | 55.9% | — |
| GDPval | 49.7% | — |
| Remote Labor Index | 2.5% | — |
| τ²-bench Airline | 83% | — |
| τ²-bench Retail | 81.6% | — |
| τ²-bench Telecom | 89.7% | — |
| DeepResearch Bench | 41.1% | — |
| PostTrainBench | — | 32% |
| GBAEval | — | 48.3% |
| GDP.pdf | — | 19% |
| LMArena Search | 1207 | — |
| METR Time Horizons | 75.3% | — |
Reasoning Kimi K3 leads
GPT-5.2: 50.2 (#35), Kimi K3: 63.0 (#17)
| Benchmark | GPT-5.2 | Kimi K3 |
|---|---|---|
| ARC-AGI-2 | 52.9% | 60.4% |
| SimpleBench | 45.8% | 60.7% |
| NYT Connections (extended) | 83.6% | 93.6% |
| ARC-AGI-1 | 86.2% | 94.5% |
| Chess Puzzles | 49% | 39% |
| LMArena Hard Prompts | 1445 | 1496 |
| Mystery Game Puzzles | 23% | 26% |
| DTBench | 90.9% | 91.2% |
| LMCA | 43.9% | 52.7% |
| Epoch Capabilities Index | 153.45 | 157.45 |
| ForecastBench | 60.1 | 61.1 |
| Kagi LLM Benchmark | 73.3% | — |
| CritPt | — | 23.4% |
| EnigmaEval | 10.4% | — |
| EBR-Bench | 23% | — |
| Surface Evolver Bench | — | 95% |
Math Kimi K3 leads
GPT-5.2: 60.0 (#38), Kimi K3: 74.2 (#16)
| Benchmark | GPT-5.2 | Kimi K3 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 67.4% | 72.2% |
| FrontierMath Tier 4 | 31.7% | 39% |
| MathArena Final-Answer Competitions | 72% | 87.8% |
| OTIS Mock AIME 2024-2025 | 96.1% | 97.2% |
| ProofBench | 15% | 87% |
| LMArena Math | 1440 | 1491 |
| FrontierMath (Feb 2025 set) | 40.7% | — |
| FrontierMath Tier 4 (v1) | 18.8% | — |
Knowledge Kimi K3 leads
GPT-5.2: 59.3 (#32), Kimi K3: 63.2 (#21)
| Benchmark | GPT-5.2 | Kimi K3 |
|---|---|---|
| GPQA Diamond | 91.4% | 93.1% |
| SimpleQA Verified | 37.1% | 50.6% |
| LMArena Expert | 1445 | 1521 |
| Humanity's Last Exam | 27.8% | — |
| Vectara Hallucination Rate | 8.4% | — |
Multimodal GPT-5.2 leads
GPT-5.2: 51.3 (#7), Kimi K3: 37.8 (#70)
| Benchmark | GPT-5.2 | Kimi K3 |
|---|---|---|
| Furniture Assembly | 38.3% | 34.2% |
| LMArena Vision | 1268 | — |
| VPCT | 84% | — |
| Blueprint-Bench 2 | — | 29.5% |
| LMArena Document | 1405 | — |
Multilingual Kimi K3 leads
GPT-5.2: 53.4 (#67), Kimi K3: 56.3 (#21)
| Benchmark | GPT-5.2 | Kimi K3 |
|---|---|---|
| LMArena Non-English | 1425 | 1466 |
| LMArena Chinese | 1460 | 1529 |
| LMArena French | 1455 | 1491 |
| LMArena German | 1448 | 1488 |
| LMArena Japanese | 1420 | 1487 |
| LMArena Korean | 1392 | 1458 |
| LMArena Russian | 1440 | 1482 |
| LMArena Spanish | 1433 | 1472 |
Instruction Following Kimi K3 leads
GPT-5.2: 74.7 (#89), Kimi K3: 77.7 (#14)
| Benchmark | GPT-5.2 | Kimi K3 |
|---|---|---|
| LMArena Instruction Following | 1417 | 1483 |
Long Context Kimi K3 leads
GPT-5.2: 44.0 (#78), Kimi K3: 45.8 (#29)
| Benchmark | GPT-5.2 | Kimi K3 |
|---|---|---|
| LMArena Longer Query | 1428 | 1494 |
| CL-bench | 18.2% | — |
Writing & Preference Kimi K3 leads
GPT-5.2: 66.8 (#32), Kimi K3: 76.6 (#4)
| Benchmark | GPT-5.2 | Kimi K3 |
|---|---|---|
| LMArena Text | 1439 | 1476 |
| LMArena Creative Writing | 1401 | 1454 |
| EQ-Bench Creative Writing | 1703 | 2082 |
| LMArena Multi-Turn | 1458 | 1488 |
| EQ-Bench 4 | — | 1339 |
Frequently asked questions
Is GPT-5.2 better than Kimi K3?
Kimi K3 is the stronger model overall, scoring 59.5 to 54.1 on the Noometry Index.
Which is cheaper, GPT-5.2 or Kimi K3?
GPT-5.2 is cheaper. It lists at $1.75 per million input tokens and $14 per million output tokens; Kimi K3 lists at $3 and $15.
Is GPT-5.2 or Kimi K3 better for coding?
Kimi K3 scores higher on coding benchmarks: 61.0 versus 51.6 in the Noometry coding category.
Which has the bigger context window?
Kimi K3 does, with 1.05M tokens against 400K.
How many benchmarks do GPT-5.2 and Kimi K3 share?
41 benchmarks have published results for both models. GPT-5.2 has 67 scored results on Noometry and Kimi K3 has 53.