Model comparison
GPT-4o vs Kimi K2.5
Kimi K2.5 is the stronger model overall, scoring 48.1 to 28.6 on the Noometry Index.
Last verified . 35 shared benchmarks.
Summary
- They share 35 benchmarks with published results for both. GPT-4o scores higher in 0 categories and Kimi K2.5 in 10 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in math, where Kimi K2.5 leads 51.8 to 10.6.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 6.4% for GPT-4o and 92.2% for Kimi K2.5.
- Kimi K2.5 is cheaper at $0.45 / $2.25 per million input/output tokens, against $2.50 / $10 for GPT-4o.
- 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-4o | Kimi K2.5 | |
|---|---|---|
| Provider | OpenAI | Moonshot AI |
| Noometry Index | 28.6 | 48.1 |
| Released | 2024-05-13 | 2026-01-27 |
| Weights | Proprietary | Open |
| Context window | 128K | 262K |
| Max output | 16K | 262K |
| Input $ / M tokens | $2.50 | $0.45 |
| Output $ / M tokens | $10 | $2.25 |
| Results tracked | 72 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K2.5 leads
GPT-4o: 24.8 (#328), Kimi K2.5: 48.8 (#53)
| Benchmark | GPT-4o | Kimi K2.5 |
|---|---|---|
| SWE-bench Verified | 31% | 73.8% |
| SWE-bench Verified (bash only) | 21.6% | 70.8% |
| WeirdML | 25.1% | 45.6% |
| LMArena Coding | 1297 | 1474 |
| Aider Polyglot | 45.3% | — |
| LMArena WebDev | — | 1437 |
| SWE-bench Multilingual | — | 67.3% |
| SciCode | — | 49% |
| GSO | 0% | — |
| BigCodeBench Instruct | 51.1% | — |
| LiveBench Coding | 51.4% | — |
| BigCodeBench Complete | 61.1% | — |
| CadEval | 26% | — |
| ALE-Bench | — | 821.65 |
| HumanEval+ | 87.2% | — |
| MBPP+ | 72.2% | — |
Agentic & Tool Use Kimi K2.5 leads
GPT-4o: 21.0 (#141), Kimi K2.5: 34.2 (#48)
| Benchmark | GPT-4o | Kimi K2.5 |
|---|---|---|
| Terminal-Bench | — | 43.2% |
| GDPval | 9.9% | — |
| TheAgentCompany | 8.6% | — |
| Cybench | 12.5% | — |
| OSWorld | — | 63.3% |
| BALROG | 32.3% | — |
| LMArena Search | 1006 | — |
| METR Time Horizons | 40.8% | — |
| Vending-Bench 2 | — | 1,198 |
Reasoning Kimi K2.5 leads
GPT-4o: 9.4 (#343), Kimi K2.5: 31.2 (#80)
| Benchmark | GPT-4o | Kimi K2.5 |
|---|---|---|
| ARC-AGI-2 | 0% | 11.8% |
| SimpleBench | 17.8% | 46.8% |
| ARC-AGI-1 | 4.5% | 65.3% |
| CritPt | 0% | 3.1% |
| Chess Puzzles | 13% | 12% |
| EnigmaEval | 0.8% | 3.4% |
| LMArena Hard Prompts | 1281 | 1453 |
| Epoch Capabilities Index | 128.97 | 148.03 |
| Kagi LLM Benchmark | — | 78.5% |
| NYT Connections (extended) | — | 69.9% |
| Thematic Generalization | — | 69.4% |
| LiveBench Reasoning | 55.8% | — |
| DTBench | 64.5% | — |
| LiveBench Data Analysis | 60.9% | — |
| LMCA | 16.6% | — |
| ForecastBench | 57.7 | — |
| LiveBench | 55.3% | — |
Math Kimi K2.5 leads
GPT-4o: 10.6 (#312), Kimi K2.5: 51.8 (#53)
| Benchmark | GPT-4o | Kimi K2.5 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 6.4% | 92.2% |
| LMArena Math | 1285 | 1470 |
| FrontierMath (Feb 2025 set) | 0.3% | 27.9% |
| FrontierMath (Tiers 1-3) | 0.4% | — |
| MathArena Final-Answer Competitions | — | 62.3% |
| Omni-MATH | 29.3% | — |
| LiveBench Math | 49.5% | — |
| MATH Level 5 | 53.3% | — |
| FrontierMath Tier 4 (v1) | — | 4.2% |
Knowledge Kimi K2.5 leads
GPT-4o: 28.8 (#242), Kimi K2.5: 53.6 (#56)
| Benchmark | GPT-4o | Kimi K2.5 |
|---|---|---|
| GPQA Diamond | 49.2% | 87.6% |
| Humanity's Last Exam | 2.7% | 24.4% |
| SimpleQA Verified | 26% | 34.3% |
| Vectara Hallucination Rate | 9.6% | 14.2% |
| LMArena Expert | 1250 | 1466 |
| MMLU-Pro | 71.3% | — |
| Confabulations | 15.3% | — |
| GPQA (HELM) | 52% | — |
| MMLU | 88.1% | — |
Multimodal Kimi K2.5 leads
GPT-4o: 34.5 (#91), Kimi K2.5: 41.1 (#39)
| Benchmark | GPT-4o | Kimi K2.5 |
|---|---|---|
| LMArena Vision | 1137 | 1269 |
| Video-MME | 71.9% | — |
| GeoBench | 71% | — |
| VPCT | 40% | — |
| LMArena Document | — | 1430 |
| ScienceQA | 88.5% | — |
Multilingual Kimi K2.5 leads
GPT-4o: 43.2 (#186), Kimi K2.5: 53.9 (#53)
| Benchmark | GPT-4o | Kimi K2.5 |
|---|---|---|
| LMArena Non-English | 1283 | 1433 |
| LMArena Chinese | 1277 | 1495 |
| LMArena French | 1304 | 1454 |
| LMArena German | 1282 | 1441 |
| LMArena Japanese | 1257 | 1421 |
| LMArena Korean | 1234 | 1410 |
| LMArena Russian | 1286 | 1435 |
| LMArena Spanish | 1292 | 1450 |
Instruction Following Kimi K2.5 leads
GPT-4o: 66.6 (#207), Kimi K2.5: 75.3 (#64)
| Benchmark | GPT-4o | Kimi K2.5 |
|---|---|---|
| LMArena Instruction Following | 1278 | 1431 |
| LiveBench Instruction Following | 68.6% | — |
| IFEval | 81.7% | — |
Long Context Kimi K2.5 leads
GPT-4o: 39.4 (#179), Kimi K2.5: 52.1 (#7)
| Benchmark | GPT-4o | Kimi K2.5 |
|---|---|---|
| Fiction.LiveBench | 66.7% | 86.1% |
| LMArena Longer Query | 1289 | 1445 |
| CL-bench | — | 19.3% |
| CL-bench Life | — | 13.2% |
Writing & Preference Kimi K2.5 leads
GPT-4o: 52.6 (#166), Kimi K2.5: 65.1 (#53)
| Benchmark | GPT-4o | Kimi K2.5 |
|---|---|---|
| LMArena Text | 1300 | 1445 |
| LMArena Creative Writing | 1292 | 1423 |
| LMArena Multi-Turn | 1302 | 1444 |
| Short-Story Creative Writing | 81.8% | — |
| EQ-Bench Creative Writing | — | 1579 |
| WildBench | 82.8% | — |
| LiveBench Language | 47.6% | — |
Frequently asked questions
Is GPT-4o better than Kimi K2.5?
Kimi K2.5 is the stronger model overall, scoring 48.1 to 28.6 on the Noometry Index.
Which is cheaper, GPT-4o 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-4o lists at $2.50 and $10.
Is GPT-4o or Kimi K2.5 better for coding?
Kimi K2.5 scores higher on coding benchmarks: 48.8 versus 24.8 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-4o and Kimi K2.5 share?
35 benchmarks have published results for both models. GPT-4o has 72 scored results on Noometry and Kimi K2.5 has 51.