Model comparison
GPT-5 Nano vs Kimi K2.5
Kimi K2.5 is the stronger model overall, scoring 48.1 to 33.5 on the Noometry Index. GPT-5 Nano costs 6.5× less per token, which makes it the better buy when Kimi K2.5's lead doesn't matter for your workload.
Last verified . 34 shared benchmarks.
Summary
- They share 34 benchmarks with published results for both. GPT-5 Nano scores higher in 0 categories and Kimi K2.5 in 10 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Kimi K2.5 leads 65.1 to 39.1.
- The biggest single-benchmark swing is ARC-AGI-1: 20.7% for GPT-5 Nano and 65.3% for Kimi K2.5.
- GPT-5 Nano is cheaper at $0.05 / $0.40 per million input/output tokens, against $0.45 / $2.25 for Kimi K2.5.
- GPT-5 Nano accepts more context: 400K tokens versus 262K.
- Kimi K2.5 has downloadable open weights; the other is API-only.
Side by side
| GPT-5 Nano | Kimi K2.5 | |
|---|---|---|
| Provider | OpenAI | Moonshot AI |
| Noometry Index | 33.5 | 48.1 |
| Released | 2025-08-07 | 2026-01-27 |
| Weights | Proprietary | Open |
| Context window | 400K | 262K |
| Max output | 128K | 262K |
| Input $ / M tokens | $0.05 | $0.45 |
| Output $ / M tokens | $0.40 | $2.25 |
| Results tracked | 49 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K2.5 leads
GPT-5 Nano: 33.6 (#254), Kimi K2.5: 48.8 (#53)
| Benchmark | GPT-5 Nano | Kimi K2.5 |
|---|---|---|
| SWE-bench Verified (bash only) | 34.8% | 70.8% |
| WeirdML | 38.1% | 45.6% |
| LMArena Coding | 1351 | 1474 |
| ALE-Bench | 718.67 | 821.65 |
| SWE-bench Verified | — | 73.8% |
| LMArena WebDev | — | 1437 |
| SWE-bench Multilingual | — | 67.3% |
| SciCode | — | 49% |
Agentic & Tool Use Kimi K2.5 leads
GPT-5 Nano: 25.8 (#106), Kimi K2.5: 34.2 (#48)
| Benchmark | GPT-5 Nano | Kimi K2.5 |
|---|---|---|
| Terminal-Bench | 21.8% | 43.2% |
| Berkeley Function Calling Leaderboard | 51.5% | — |
| OSWorld | — | 63.3% |
| Vending-Bench 2 | — | 1,198 |
Reasoning Kimi K2.5 leads
GPT-5 Nano: 16.3 (#306), Kimi K2.5: 31.2 (#80)
| Benchmark | GPT-5 Nano | Kimi K2.5 |
|---|---|---|
| ARC-AGI-2 | 2.6% | 11.8% |
| Kagi LLM Benchmark | 62.2% | 78.5% |
| ARC-AGI-1 | 20.7% | 65.3% |
| Chess Puzzles | 27% | 12% |
| LMArena Hard Prompts | 1328 | 1453 |
| Epoch Capabilities Index | 139.38 | 148.03 |
| SimpleBench | — | 46.8% |
| NYT Connections (extended) | — | 69.9% |
| CritPt | — | 3.1% |
| EnigmaEval | — | 3.4% |
| Thematic Generalization | — | 69.4% |
| Mystery Game Puzzles | 9% | — |
| DTBench | 62.7% | — |
| LMCA | 7.9% | — |
| ForecastBench | 59.1 | — |
Math Kimi K2.5 leads
GPT-5 Nano: 29.4 (#241), Kimi K2.5: 51.8 (#53)
| Benchmark | GPT-5 Nano | Kimi K2.5 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 81.1% | 92.2% |
| LMArena Math | 1317 | 1470 |
| FrontierMath (Feb 2025 set) | 8.3% | 27.9% |
| FrontierMath Tier 4 (v1) | 2.1% | 4.2% |
| FrontierMath (Tiers 1-3) | 20% | — |
| FrontierMath Tier 4 | 2.4% | — |
| MathArena Final-Answer Competitions | — | 62.3% |
| ProofBench | 12% | — |
| Omni-MATH | 54.6% | — |
| MATH Level 5 | 95.2% | — |
Knowledge Kimi K2.5 leads
GPT-5 Nano: 35.9 (#178), Kimi K2.5: 53.6 (#56)
| Benchmark | GPT-5 Nano | Kimi K2.5 |
|---|---|---|
| GPQA Diamond | 69.4% | 87.6% |
| SimpleQA Verified | 11.7% | 34.3% |
| Vectara Hallucination Rate | 10.5% | 14.2% |
| LMArena Expert | 1321 | 1466 |
| Humanity's Last Exam | — | 24.4% |
| MMLU-Pro | 77.8% | — |
| GPQA (HELM) | 67.9% | — |
Multimodal Kimi K2.5 leads
GPT-5 Nano: 31.3 (#108), Kimi K2.5: 41.1 (#39)
| Benchmark | GPT-5 Nano | Kimi K2.5 |
|---|---|---|
| LMArena Vision | 1159 | 1269 |
| VPCT | 37.2% | — |
| LMArena Document | — | 1430 |
Multilingual Kimi K2.5 leads
GPT-5 Nano: 45.3 (#172), Kimi K2.5: 53.9 (#53)
| Benchmark | GPT-5 Nano | Kimi K2.5 |
|---|---|---|
| LMArena Non-English | 1313 | 1433 |
| LMArena Chinese | 1356 | 1495 |
| LMArena German | 1327 | 1441 |
| LMArena Japanese | 1226 | 1421 |
| LMArena Korean | 1269 | 1410 |
| LMArena Russian | 1296 | 1435 |
| LMArena Spanish | 1360 | 1450 |
| LMArena French | — | 1454 |
Instruction Following Too close to call
GPT-5 Nano: 75.0 (#79), Kimi K2.5: 75.3 (#64)
| Benchmark | GPT-5 Nano | Kimi K2.5 |
|---|---|---|
| LMArena Instruction Following | 1306 | 1431 |
| IFEval | 93.2% | — |
Long Context Kimi K2.5 leads
GPT-5 Nano: 31.3 (#281), Kimi K2.5: 52.1 (#7)
| Benchmark | GPT-5 Nano | Kimi K2.5 |
|---|---|---|
| Fiction.LiveBench | 44.4% | 86.1% |
| LMArena Longer Query | 1312 | 1445 |
| CL-bench | — | 19.3% |
| CL-bench Life | — | 13.2% |
Writing & Preference Kimi K2.5 leads
GPT-5 Nano: 39.1 (#249), Kimi K2.5: 65.1 (#53)
| Benchmark | GPT-5 Nano | Kimi K2.5 |
|---|---|---|
| LMArena Text | 1320 | 1445 |
| LMArena Creative Writing | 1249 | 1423 |
| EQ-Bench Creative Writing | 705 | 1579 |
| LMArena Multi-Turn | 1311 | 1444 |
| WildBench | 80.6% | — |
Frequently asked questions
Is GPT-5 Nano better than Kimi K2.5?
Kimi K2.5 is the stronger model overall, scoring 48.1 to 33.5 on the Noometry Index. GPT-5 Nano costs 6.5× less per token, which makes it the better buy when Kimi K2.5's lead doesn't matter for your workload.
Which is cheaper, GPT-5 Nano or Kimi K2.5?
GPT-5 Nano is cheaper. It lists at $0.05 per million input tokens and $0.40 per million output tokens; Kimi K2.5 lists at $0.45 and $2.25.
Is GPT-5 Nano or Kimi K2.5 better for coding?
Kimi K2.5 scores higher on coding benchmarks: 48.8 versus 33.6 in the Noometry coding category.
Which has the bigger context window?
GPT-5 Nano does, with 400K tokens against 262K.
How many benchmarks do GPT-5 Nano and Kimi K2.5 share?
34 benchmarks have published results for both models. GPT-5 Nano has 49 scored results on Noometry and Kimi K2.5 has 51.