Model comparison
GPT-4 Turbo vs Kimi K2.5
Kimi K2.5 is the stronger model overall, scoring 47.8 to 30.3 on the Noometry Index.
Last verified . 24 shared benchmarks.
Summary
- They share 24 benchmarks with published results for both. GPT-4 Turbo scores higher in 0 categories and Kimi K2.5 in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where Kimi K2.5 leads 50.3 to 8.6.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 6.7% for GPT-4 Turbo and 92.2% for Kimi K2.5.
- Kimi K2.5 is cheaper at $0.45 / $2.25 per million input/output tokens, against $10 / $30 for GPT-4 Turbo.
- Kimi K2.5 accepts more context: 262K tokens versus 128K.
- Kimi K2.5 has downloadable open weights; the other is API-only.
Side by side
| GPT-4 Turbo | Kimi K2.5 | |
|---|---|---|
| Provider | OpenAI | Moonshot AI |
| Noometry Index | 30.3 | 47.8 |
| Released | 2023-11-06 | 2026-01-27 |
| Weights | Proprietary | Open |
| Context window | 128K | 262K |
| Max output | 4K | 262K |
| Input $ / M tokens | $10 | $0.45 |
| Output $ / M tokens | $30 | $2.25 |
| Results tracked | 36 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K2.5 leads
GPT-4 Turbo: 33.4 (#247), Kimi K2.5: 48.7 (#48)
| Benchmark | GPT-4 Turbo | Kimi K2.5 |
|---|---|---|
| WeirdML | 18% | 45.6% |
| LMArena Coding | 1268 | 1474 |
| SWE-bench Verified | — | 73.8% |
| SWE-bench Verified (bash only) | — | 70.8% |
| LMArena WebDev | — | 1436 |
| SWE-bench Multilingual | — | 67.3% |
| SciCode | — | 49% |
| BigCodeBench Instruct | 48.2% | — |
| BigCodeBench Complete | 58.2% | — |
| ALE-Bench | — | 821.65 |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73.3% | — |
Agentic & Tool Use Not comparable
GPT-4 Turbo: —, Kimi K2.5: 34.0 (#44)
| Benchmark | GPT-4 Turbo | Kimi K2.5 |
|---|---|---|
| Terminal-Bench | — | 43.2% |
| OSWorld | — | 63.3% |
| METR Time Horizons | 36.7% | — |
| Vending-Bench 2 | — | 1,198 |
Reasoning Kimi K2.5 leads
GPT-4 Turbo: 14.9 (#311), Kimi K2.5: 31.2 (#74)
| Benchmark | GPT-4 Turbo | Kimi K2.5 |
|---|---|---|
| SimpleBench | 25.1% | 46.8% |
| Chess Puzzles | 6% | 12% |
| LMArena Hard Prompts | 1251 | 1454 |
| Epoch Capabilities Index | 127.25 | 148.05 |
| ARC-AGI-2 | — | 11.8% |
| Kagi LLM Benchmark | — | 78.5% |
| NYT Connections (extended) | — | 69.9% |
| ARC-AGI-1 | — | 65.3% |
| CritPt | — | 3.1% |
| EnigmaEval | — | 3.4% |
| Thematic Generalization | — | 69.4% |
| DTBench | 61.6% | — |
| LMCA | 9.8% | — |
| ForecastBench | 59.4 | — |
Math Kimi K2.5 leads
GPT-4 Turbo: 8.6 (#316), Kimi K2.5: 50.3 (#46)
| Benchmark | GPT-4 Turbo | Kimi K2.5 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 6.7% | 92.2% |
| LMArena Math | 1272 | 1470 |
| FrontierMath (Tiers 1-3) | 0.7% | — |
| MathArena Final-Answer Competitions | — | 62.3% |
| MATH Level 5 | 46.7% | — |
| FrontierMath (Feb 2025 set) | — | 27.9% |
| FrontierMath Tier 4 (v1) | — | 4.2% |
Knowledge Kimi K2.5 leads
GPT-4 Turbo: 24.3 (#263), Kimi K2.5: 53.6 (#52)
| Benchmark | GPT-4 Turbo | Kimi K2.5 |
|---|---|---|
| GPQA Diamond | 46.6% | 87.6% |
| LMArena Expert | 1223 | 1466 |
| Humanity's Last Exam | — | 24.4% |
| SimpleQA Verified | — | 34.3% |
| Confabulations | 28.4% | — |
| Vectara Hallucination Rate | — | 14.2% |
| MMLU | 81.3% | — |
Multimodal Kimi K2.5 leads
GPT-4 Turbo: 30.0 (#104), Kimi K2.5: 40.4 (#37)
| Benchmark | GPT-4 Turbo | Kimi K2.5 |
|---|---|---|
| LMArena Vision | 1090 | 1269 |
| LMArena Document | — | 1430 |
Multilingual Kimi K2.5 leads
GPT-4 Turbo: 40.7 (#211), Kimi K2.5: 54.1 (#50)
| Benchmark | GPT-4 Turbo | Kimi K2.5 |
|---|---|---|
| LMArena Non-English | 1245 | 1433 |
| LMArena Chinese | 1242 | 1495 |
| LMArena French | 1275 | 1454 |
| LMArena German | 1259 | 1441 |
| LMArena Japanese | 1194 | 1422 |
| LMArena Korean | 1187 | 1412 |
| LMArena Russian | 1259 | 1436 |
| LMArena Spanish | 1260 | 1450 |
Instruction Following Kimi K2.5 leads
GPT-4 Turbo: 66.0 (#211), Kimi K2.5: 75.5 (#60)
| Benchmark | GPT-4 Turbo | Kimi K2.5 |
|---|---|---|
| LMArena Instruction Following | 1250 | 1431 |
Long Context Kimi K2.5 leads
GPT-4 Turbo: 38.0 (#200), Kimi K2.5: 52.1 (#7)
| Benchmark | GPT-4 Turbo | Kimi K2.5 |
|---|---|---|
| LMArena Longer Query | 1254 | 1445 |
| Fiction.LiveBench | — | 86.1% |
| CL-bench | — | 19.3% |
| CL-bench Life | — | 13.2% |
Writing & Preference Kimi K2.5 leads
GPT-4 Turbo: 47.9 (#201), Kimi K2.5: 65.3 (#51)
| Benchmark | GPT-4 Turbo | Kimi K2.5 |
|---|---|---|
| LMArena Text | 1272 | 1445 |
| LMArena Creative Writing | 1269 | 1423 |
| LMArena Multi-Turn | 1267 | 1444 |
| EQ-Bench Creative Writing | — | 1579 |
Frequently asked questions
Is GPT-4 Turbo better than Kimi K2.5?
Kimi K2.5 is the stronger model overall, scoring 47.8 to 30.3 on the Noometry Index.
Which is cheaper, GPT-4 Turbo or Kimi K2.5?
Kimi K2.5 is cheaper. It lists at $0.45 per million input tokens and $2.25 per million output tokens; GPT-4 Turbo lists at $10 and $30.
Is GPT-4 Turbo or Kimi K2.5 better for coding?
Kimi K2.5 scores higher on coding benchmarks: 48.7 versus 33.4 in the Noometry coding category.
Which has the bigger context window?
Kimi K2.5 does, with 262K tokens against 128K.
How many benchmarks do GPT-4 Turbo and Kimi K2.5 share?
24 benchmarks have published results for both models. GPT-4 Turbo has 36 scored results on Noometry and Kimi K2.5 has 51.