Model comparison
GPT-4 vs Kimi K2.6
Kimi K2.6 is the stronger model overall, scoring 47.7 to 29.1 on the Noometry Index.
Last verified . 26 shared benchmarks.
Summary
- They share 26 benchmarks with published results for both. GPT-4 scores higher in 0 categories and Kimi K2.6 in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Kimi K2.6 leads 57.0 to 10.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 1.1% for GPT-4 and 96.1% for Kimi K2.6.
- Kimi K2.6 is cheaper at $0.95 / $4 per million input/output tokens, against $30 / $60 for GPT-4.
- Kimi K2.6 accepts more context: 262K tokens versus 8K.
- Kimi K2.6 has downloadable open weights; the other is API-only.
Side by side
| GPT-4 | Kimi K2.6 | |
|---|---|---|
| Provider | OpenAI | Moonshot AI |
| Noometry Index | 29.1 | 47.7 |
| Released | 2023-03-14 | 2026-04-20 |
| Weights | Proprietary | Open |
| Context window | 8K | 262K |
| Max output | 8K | 262K |
| Input $ / M tokens | $30 | $0.95 |
| Output $ / M tokens | $60 | $4 |
| Results tracked | 38 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K2.6 leads
GPT-4: 31.6 (#283), Kimi K2.6: 50.7 (#43)
| Benchmark | GPT-4 | Kimi K2.6 |
|---|---|---|
| WeirdML | 12.4% | 55.9% |
| LMArena Coding | 1254 | 1488 |
| SWE-bench Verified | — | 76.7% |
| LMArena WebDev | — | 1509 |
| SciCode | — | 53.5% |
| BigCodeBench Instruct | 46% | — |
| BigCodeBench Complete | 57.2% | — |
| ALE-Bench | — | 1,093 |
| HumanEval+ | 79.3% | — |
Agentic & Tool Use Not comparable
GPT-4: —, Kimi K2.6: 21.9 (#137)
| Benchmark | GPT-4 | Kimi K2.6 |
|---|---|---|
| OSWorld 2.0 | — | 4.6% |
| ExploitBench | — | 18.4% |
| GBAEval | — | 0.9% |
| GDP.pdf | — | 12% |
| METR Time Horizons | 36.1% | — |
| Vending-Bench 2 | — | 6,205 |
Reasoning Kimi K2.6 leads
GPT-4: 17.8 (#289), Kimi K2.6: 40.5 (#55)
| Benchmark | GPT-4 | Kimi K2.6 |
|---|---|---|
| Chess Puzzles | 4% | 26% |
| LMArena Hard Prompts | 1241 | 1470 |
| Mystery Game Puzzles | 12% | 18% |
| DTBench | 62.7% | 90.9% |
| LMCA | 17.1% | 37.3% |
| Epoch Capabilities Index | 125.89 | 151.05 |
| NYT Connections (extended) | — | 87.2% |
| CritPt | — | 8% |
| EBR-Bench | — | 2.4% |
| BIG-Bench Hard | 75.1% | — |
| ForecastBench | 57.8 | — |
| HellaSwag | 95.3% | — |
| WinoGrande | 87.5% | — |
Math Kimi K2.6 leads
GPT-4: 10.8 (#309), Kimi K2.6: 57.0 (#41)
| Benchmark | GPT-4 | Kimi K2.6 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 1.1% | 96.1% |
| LMArena Math | 1269 | 1475 |
| FrontierMath (Tiers 1-3) | — | 57.2% |
| FrontierMath Tier 4 | — | 25.6% |
| MathArena Final-Answer Competitions | — | 72.9% |
| ProofBench | — | 16% |
| MATH Level 5 | 23% | — |
| FrontierMath (Feb 2025 set) | — | 39% |
| FrontierMath Tier 4 (v1) | — | 14.6% |
| GSM8K | 92% | — |
Knowledge Kimi K2.6 leads
GPT-4: 18.4 (#282), Kimi K2.6: 54.0 (#54)
| Benchmark | GPT-4 | Kimi K2.6 |
|---|---|---|
| GPQA Diamond | 35.7% | 90.8% |
| LMArena Expert | 1211 | 1491 |
| SimpleQA Verified | — | 34.9% |
| Vectara Hallucination Rate | — | 10.8% |
| MMLU | 86.4% | — |
| TriviaQA | 84.8% | — |
Multimodal Not comparable
GPT-4: —, Kimi K2.6: 31.6 (#103)
| Benchmark | GPT-4 | Kimi K2.6 |
|---|---|---|
| LMArena Vision | — | 1283 |
| Blueprint-Bench 2 | — | 3.9% |
| Furniture Assembly | — | 21.7% |
| LMArena Document | — | 1451 |
Multilingual Kimi K2.6 leads
GPT-4: 40.6 (#215), Kimi K2.6: 54.9 (#37)
| Benchmark | GPT-4 | Kimi K2.6 |
|---|---|---|
| LMArena Non-English | 1246 | 1446 |
| LMArena Chinese | 1242 | 1521 |
| LMArena French | 1283 | 1471 |
| LMArena German | 1251 | 1450 |
| LMArena Japanese | 1209 | 1443 |
| LMArena Korean | 1184 | 1427 |
| LMArena Russian | 1251 | 1446 |
| LMArena Spanish | 1261 | 1464 |
Instruction Following Kimi K2.6 leads
GPT-4: 65.3 (#222), Kimi K2.6: 76.3 (#43)
| Benchmark | GPT-4 | Kimi K2.6 |
|---|---|---|
| LMArena Instruction Following | 1241 | 1451 |
Long Context Kimi K2.6 leads
GPT-4: 37.7 (#212), Kimi K2.6: 44.9 (#52)
| Benchmark | GPT-4 | Kimi K2.6 |
|---|---|---|
| LMArena Longer Query | 1244 | 1468 |
Writing & Preference Kimi K2.6 leads
GPT-4: 34.9 (#268), Kimi K2.6: 68.5 (#26)
| Benchmark | GPT-4 | Kimi K2.6 |
|---|---|---|
| LMArena Text | 1263 | 1455 |
| LMArena Creative Writing | 1244 | 1434 |
| EQ-Bench Creative Writing | 752 | 1725 |
| LMArena Multi-Turn | 1257 | 1453 |
| EQ-Bench 4 | — | 1202 |
Frequently asked questions
Is GPT-4 better than Kimi K2.6?
Kimi K2.6 is the stronger model overall, scoring 47.7 to 29.1 on the Noometry Index.
Which is cheaper, GPT-4 or Kimi K2.6?
Kimi K2.6 is cheaper. It lists at $0.95 per million input tokens and $4 per million output tokens; GPT-4 lists at $30 and $60.
Is GPT-4 or Kimi K2.6 better for coding?
Kimi K2.6 scores higher on coding benchmarks: 50.7 versus 31.6 in the Noometry coding category.
Which has the bigger context window?
Kimi K2.6 does, with 262K tokens against 8K.
How many benchmarks do GPT-4 and Kimi K2.6 share?
26 benchmarks have published results for both models. GPT-4 has 38 scored results on Noometry and Kimi K2.6 has 51.