Model comparison
GPT-3.5-turbo vs Kimi K3
Kimi K3 is the stronger model overall, scoring 59.5 to 23.1 on the Noometry Index. GPT-3.5-turbo costs 8.0× less per token, which makes it the better buy when Kimi K3's lead doesn't matter for your workload.
Last verified . 28 shared benchmarks.
Summary
- They share 28 benchmarks with published results for both. GPT-3.5-turbo scores higher in 0 categories and Kimi K3 in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Kimi K3 leads 74.0 to 6.0.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 2.2% for GPT-3.5-turbo and 97.2% for Kimi K3.
- GPT-3.5-turbo is cheaper at $0.50 / $1.50 per million input/output tokens, against $3 / $15 for Kimi K3.
- Kimi K3 accepts more context: 1.05M tokens versus 16K.
- Kimi K3 has downloadable open weights; the other is API-only.
Side by side
| GPT-3.5-turbo | Kimi K3 | |
|---|---|---|
| Provider | OpenAI | Moonshot AI |
| Noometry Index | 23.1 | 59.5 |
| Released | 2023-03-01 | 2026-07-16 |
| Weights | Proprietary | Open |
| Context window | 16K | 1.05M |
| Max output | 4K | 1.05M |
| Input $ / M tokens | $0.50 | $3 |
| Output $ / M tokens | $1.50 | $15 |
| Results tracked | 44 | 53 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K3 leads
GPT-3.5-turbo: 23.5 (#326), Kimi K3: 60.6 (#8)
| Benchmark | GPT-3.5-turbo | Kimi K3 |
|---|---|---|
| WeirdML | 3.5% | 82.6% |
| LMArena Coding | 1137 | 1509 |
| DeepSWE | — | 68.5% |
| FrontierCode | — | 44.2% |
| LMArena WebDev | — | 1659 |
| FrontierSWE | — | 25.9% |
| SciCode | — | 59.5% |
| BigCodeBench Instruct | 39.1% | — |
| BigCodeBench Complete | 50.6% | — |
| ALE-Bench | — | 1,524 |
| HumanEval+ | 70.7% | — |
| MBPP+ | 69.7% | — |
Agentic & Tool Use Not comparable
GPT-3.5-turbo: —, Kimi K3: 41.8 (#19)
| Benchmark | GPT-3.5-turbo | Kimi K3 |
|---|---|---|
| APEX-Agents | — | 50.6% |
| τ²-bench Banking | — | 37.1% |
| PostTrainBench | — | 32% |
| GBAEval | — | 48.3% |
| GDP.pdf | — | 19% |
| METR Time Horizons | 21.5% | — |
| Vending-Bench 2 | — | 5,165 |
Reasoning Kimi K3 leads
GPT-3.5-turbo: 13.3 (#328), Kimi K3: 63.0 (#15)
| Benchmark | GPT-3.5-turbo | Kimi K3 |
|---|---|---|
| Chess Puzzles | 0% | 39% |
| LMArena Hard Prompts | 1108 | 1495 |
| Mystery Game Puzzles | 3% | 26% |
| DTBench | 48.5% | 91.2% |
| LMCA | 9.7% | 52.7% |
| Epoch Capabilities Index | 118.48 | 157.68 |
| ForecastBench | 50.4 | 61.1 |
| ARC-AGI-2 | — | 60.4% |
| SimpleBench | — | 60.7% |
| NYT Connections (extended) | — | 93.6% |
| ARC-AGI-1 | — | 94.5% |
| CritPt | — | 23.4% |
| Surface Evolver Bench | — | 95% |
| Adversarial NLI | 58.1% | — |
| BIG-Bench Hard | 61.6% | — |
| CommonsenseQA 2.0 | 57% | — |
| WinoGrande | 81.6% | — |
Math Kimi K3 leads
GPT-3.5-turbo: 6.0 (#321), Kimi K3: 74.0 (#13)
| Benchmark | GPT-3.5-turbo | Kimi K3 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 0% | 72.2% |
| OTIS Mock AIME 2024-2025 | 2.2% | 97.2% |
| LMArena Math | 1142 | 1493 |
| FrontierMath Tier 4 | — | 39% |
| MathArena Final-Answer Competitions | — | 87.8% |
| ProofBench | — | 87% |
| MATH Level 5 | 15.9% | — |
| GSM8K | 57.8% | — |
Knowledge Kimi K3 leads
GPT-3.5-turbo: 10.0 (#298), Kimi K3: 63.2 (#19)
| Benchmark | GPT-3.5-turbo | Kimi K3 |
|---|---|---|
| GPQA Diamond | 28% | 93.1% |
| LMArena Expert | 1070 | 1520 |
| SimpleQA Verified | — | 50.6% |
| ARC (AI2) Challenge | 87.4% | — |
| BoolQ | 87% | — |
| MMLU | 71.4% | — |
| OpenBookQA | 86% | — |
| TriviaQA | 85.8% | — |
Multimodal Not comparable
GPT-3.5-turbo: —, Kimi K3: 38.4 (#58)
| Benchmark | GPT-3.5-turbo | Kimi K3 |
|---|---|---|
| Blueprint-Bench 2 | — | 29.5% |
| Furniture Assembly | — | 34.2% |
Multilingual Kimi K3 leads
GPT-3.5-turbo: 31.6 (#253), Kimi K3: 56.5 (#20)
| Benchmark | GPT-3.5-turbo | Kimi K3 |
|---|---|---|
| LMArena Non-English | 1108 | 1466 |
| LMArena Chinese | 1075 | 1528 |
| LMArena French | 1117 | 1488 |
| LMArena German | 1090 | 1474 |
| LMArena Japanese | 1043 | 1492 |
| LMArena Korean | 1019 | 1465 |
| LMArena Russian | 1123 | 1481 |
| LMArena Spanish | 1120 | 1478 |
Instruction Following Kimi K3 leads
GPT-3.5-turbo: 58.1 (#257), Kimi K3: 77.8 (#13)
| Benchmark | GPT-3.5-turbo | Kimi K3 |
|---|---|---|
| LMArena Instruction Following | 1119 | 1481 |
Long Context Kimi K3 leads
GPT-3.5-turbo: 34.0 (#249), Kimi K3: 45.8 (#27)
| Benchmark | GPT-3.5-turbo | Kimi K3 |
|---|---|---|
| LMArena Longer Query | 1121 | 1494 |
Writing & Preference Kimi K3 leads
GPT-3.5-turbo: 25.4 (#300), Kimi K3: 76.7 (#4)
| Benchmark | GPT-3.5-turbo | Kimi K3 |
|---|---|---|
| LMArena Text | 1125 | 1476 |
| LMArena Creative Writing | 1092 | 1452 |
| EQ-Bench Creative Writing | 451 | 2082 |
| LMArena Multi-Turn | 1117 | 1488 |
| EQ-Bench 4 | — | 1339 |
Frequently asked questions
Is GPT-3.5-turbo better than Kimi K3?
Kimi K3 is the stronger model overall, scoring 59.5 to 23.1 on the Noometry Index. GPT-3.5-turbo costs 8.0× less per token, which makes it the better buy when Kimi K3's lead doesn't matter for your workload.
Which is cheaper, GPT-3.5-turbo or Kimi K3?
GPT-3.5-turbo is cheaper. It lists at $0.50 per million input tokens and $1.50 per million output tokens; Kimi K3 lists at $3 and $15.
Is GPT-3.5-turbo or Kimi K3 better for coding?
Kimi K3 scores higher on coding benchmarks: 60.6 versus 23.5 in the Noometry coding category.
Which has the bigger context window?
Kimi K3 does, with 1.05M tokens against 16K.
How many benchmarks do GPT-3.5-turbo and Kimi K3 share?
28 benchmarks have published results for both models. GPT-3.5-turbo has 44 scored results on Noometry and Kimi K3 has 53.