# GPT-5.5 vs Kimi K2.5

> GPT-5.5 is the stronger model overall, scoring 63.4 to 47.8 on the Noometry Index. Kimi K2.5 costs 13× less per token, which makes it the better buy when GPT-5.5's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/gpt-5-5-vs-kimi-k2-5
- Last updated: 2026-10-10
- Shared benchmarks: 43

## Summary

- They share 43 benchmarks with published results for both. GPT-5.5 scores higher in 9 categories and Kimi K2.5 in 1 category; 10 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GPT-5.5 leads 72.7 to 31.2.
- The biggest single-benchmark swing is ARC-AGI-2: 85% for GPT-5.5 and 11.8% for Kimi K2.5.
- Kimi K2.5 is cheaper at $0.45 / $2.25 per million input/output tokens, against $5 / $30 for GPT-5.5.
- GPT-5.5 accepts more context: 1.05M tokens versus 262K.
- Kimi K2.5 has downloadable open weights; the other is API-only.

## Snapshot

| | GPT-5.5 | Kimi K2.5 |
|---|---|---|
| Provider | OpenAI | Moonshot AI |
| Noometry Index | 63.4 | 47.8 |
| Rank | 8 | 50 |
| Context | 1.05M | 262K |
| Input $/M | $5 | $0.45 |
| Output $/M | $30 | $2.25 |
| Weights | Proprietary | Open |

## Coding

- GPT-5.5: 58.3 (#13)
- Kimi K2.5: 48.7 (#48)

| Benchmark | GPT-5.5 | Kimi K2.5 |
|---|---|---|
| SWE-bench Verified | 80.6% | 73.8% |
| LMArena WebDev | 1512 | 1436 |
| SciCode | 56.1% | 49% |
| WeirdML | 84.9% | 45.6% |
| LMArena Coding | 1494 | 1474 |
| ALE-Bench | 1,943 | 821.65 |
| DeepSWE | 67% | — |
| FrontierCode | 43% | — |
| SWE-bench Verified (bash only) | — | 70.8% |
| SWE-bench Multilingual | — | 67.3% |
| GSO | 40.2% | — |
| MirrorCode | 10% | — |

## Agentic & Tool Use

- GPT-5.5: 50.7 (#6)
- Kimi K2.5: 34.0 (#44)

| Benchmark | GPT-5.5 | Kimi K2.5 |
|---|---|---|
| Terminal-Bench | 84.7% | 43.2% |
| Vending-Bench 2 | 7,524 | 1,198 |
| APEX-Agents | 55.1% | — |
| OSWorld 2.0 | 13% | — |
| Remote Labor Index | 6.3% | — |
| τ²-bench Banking | 44.6% | — |
| DeepResearch Bench | 54% | — |
| OSWorld | — | 63.3% |
| PostTrainBench | 27.2% | — |
| ExploitBench | 47.4% | — |
| GBAEval | 53.2% | — |
| GDP.pdf | 26% | — |
| LMArena Search | 1242 | — |

## Reasoning

- GPT-5.5: 72.7 (#8)
- Kimi K2.5: 31.2 (#74)

| Benchmark | GPT-5.5 | Kimi K2.5 |
|---|---|---|
| ARC-AGI-2 | 85% | 11.8% |
| SimpleBench | 69% | 46.8% |
| Kagi LLM Benchmark | 88.8% | 78.5% |
| NYT Connections (extended) | 96.2% | 69.9% |
| ARC-AGI-1 | 95% | 65.3% |
| CritPt | 27.1% | 3.1% |
| Chess Puzzles | 54% | 12% |
| LMArena Hard Prompts | 1489 | 1454 |
| Epoch Capabilities Index | 159.25 | 148.05 |
| EnigmaEval | — | 3.4% |
| Thematic Generalization | — | 69.4% |
| EBR-Bench | 34.3% | — |
| Mystery Game Puzzles | 56% | — |
| DTBench | 96% | — |
| LMCA | 54.3% | — |
| Surface Evolver Bench | 88.1% | — |
| Bench to the Future 3 | 0.14 | — |
| ForecastBench | 60.6 | — |

## Math

- GPT-5.5: 81.5 (#9)
- Kimi K2.5: 50.3 (#46)

| Benchmark | GPT-5.5 | Kimi K2.5 |
|---|---|---|
| MathArena Final-Answer Competitions | 94.3% | 62.3% |
| OTIS Mock AIME 2024-2025 | 100% | 92.2% |
| LMArena Math | 1486 | 1470 |
| FrontierMath (Feb 2025 set) | 51.7% | 27.9% |
| FrontierMath Tier 4 (v1) | 35.4% | 4.2% |
| FrontierMath (Tiers 1-3) | 85.3% | — |
| FrontierMath Tier 4 | 72.5% | — |
| ProofBench | 50% | — |
| FrontierMath Erdős | 0% | — |

## Knowledge

- GPT-5.5: 64.4 (#14)
- Kimi K2.5: 53.6 (#52)

| Benchmark | GPT-5.5 | Kimi K2.5 |
|---|---|---|
| GPQA Diamond | 94% | 87.6% |
| SimpleQA Verified | 63% | 34.3% |
| Vectara Hallucination Rate | 9.3% | 14.2% |
| LMArena Expert | 1508 | 1466 |
| Humanity's Last Exam | — | 24.4% |

## Multimodal

- GPT-5.5: 47.1 (#11)
- Kimi K2.5: 40.4 (#37)

| Benchmark | GPT-5.5 | Kimi K2.5 |
|---|---|---|
| LMArena Vision | 1297 | 1269 |
| LMArena Document | 1486 | 1430 |
| Blueprint-Bench 2 | 36.2% | — |
| Furniture Assembly | 44.2% | — |

## Multilingual

- GPT-5.5: 56.5 (#19)
- Kimi K2.5: 54.1 (#50)

| Benchmark | GPT-5.5 | Kimi K2.5 |
|---|---|---|
| LMArena Non-English | 1467 | 1433 |
| LMArena Chinese | 1535 | 1495 |
| LMArena French | 1486 | 1454 |
| LMArena German | 1478 | 1441 |
| LMArena Japanese | 1498 | 1422 |
| LMArena Korean | 1461 | 1412 |
| LMArena Russian | 1473 | 1436 |
| LMArena Spanish | 1469 | 1450 |

## Instruction Following

- GPT-5.5: 77.7 (#18)
- Kimi K2.5: 75.5 (#60)

| Benchmark | GPT-5.5 | Kimi K2.5 |
|---|---|---|
| LMArena Instruction Following | 1479 | 1431 |

## Long Context

- GPT-5.5: 48.3 (#12)
- Kimi K2.5: 52.1 (#7)

| Benchmark | GPT-5.5 | Kimi K2.5 |
|---|---|---|
| CL-bench Life | 22.2% | 13.2% |
| LMArena Longer Query | 1484 | 1445 |
| Fiction.LiveBench | — | 86.1% |
| CL-bench | — | 19.3% |

## Writing & Preference

- GPT-5.5: 72.8 (#13)
- Kimi K2.5: 65.3 (#51)

| Benchmark | GPT-5.5 | Kimi K2.5 |
|---|---|---|
| LMArena Text | 1471 | 1445 |
| LMArena Creative Writing | 1455 | 1423 |
| EQ-Bench Creative Writing | 1844 | 1579 |
| LMArena Multi-Turn | 1477 | 1444 |
| EQ-Bench 4 | 1315 | — |

## FAQ

### Is GPT-5.5 better than Kimi K2.5?

GPT-5.5 is the stronger model overall, scoring 63.4 to 47.8 on the Noometry Index. Kimi K2.5 costs 13× less per token, which makes it the better buy when GPT-5.5's lead doesn't matter for your workload.

### Which is cheaper, GPT-5.5 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-5.5 lists at $5 and $30.

### Is GPT-5.5 or Kimi K2.5 better for coding?

GPT-5.5 scores higher on coding benchmarks: 58.3 versus 48.7 in the Noometry coding category.

### Which has the bigger context window?

GPT-5.5 does, with 1.05M tokens against 262K.

### How many benchmarks do GPT-5.5 and Kimi K2.5 share?

43 benchmarks have published results for both models. GPT-5.5 has 71 scored results on Noometry and Kimi K2.5 has 51.