Model comparison
GPT-5 Mini vs Kimi K2.6
Kimi K2.6 is the stronger model overall, scoring 47.7 to 41.8 on the Noometry Index. GPT-5 Mini costs 2.5× less per token, which makes it the better buy when Kimi K2.6's lead doesn't matter for your workload.
Last verified . 39 shared benchmarks.
Summary
- They share 39 benchmarks with published results for both. GPT-5 Mini scores higher in 2 categories and Kimi K2.6 in 8 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Kimi K2.6 leads 40.5 to 23.9.
- The biggest single-benchmark swing is GPQA Diamond: 75% for GPT-5 Mini and 90.8% for Kimi K2.6.
- GPT-5 Mini is cheaper at $0.25 / $2 per million input/output tokens, against $0.95 / $4 for Kimi K2.6.
- GPT-5 Mini accepts more context: 400K tokens versus 262K.
- Kimi K2.6 has downloadable open weights; the other is API-only.
Side by side
| GPT-5 Mini | Kimi K2.6 | |
|---|---|---|
| Provider | OpenAI | Moonshot AI |
| Noometry Index | 41.8 | 47.7 |
| Released | 2025-08-07 | 2026-04-20 |
| Weights | Proprietary | Open |
| Context window | 400K | 262K |
| Max output | 128K | 262K |
| Input $ / M tokens | $0.25 | $0.95 |
| Output $ / M tokens | $2 | $4 |
| Results tracked | 60 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K2.6 leads
GPT-5 Mini: 40.1 (#146), Kimi K2.6: 50.7 (#43)
| Benchmark | GPT-5 Mini | Kimi K2.6 |
|---|---|---|
| SWE-bench Verified | 64.7% | 76.7% |
| SciCode | 39.2% | 53.5% |
| WeirdML | 52.7% | 55.9% |
| LMArena Coding | 1406 | 1488 |
| ALE-Bench | 799.77 | 1,093 |
| SWE-bench Verified (bash only) | 59.8% | — |
| LMArena WebDev | — | 1509 |
| SWE-bench Multilingual | 39.7% | — |
| AlgoTune | 1.38 | — |
Agentic & Tool Use GPT-5 Mini leads
GPT-5 Mini: 31.1 (#70), Kimi K2.6: 21.9 (#137)
| Benchmark | GPT-5 Mini | Kimi K2.6 |
|---|---|---|
| Vending-Bench 2 | -31.18 | 6,205 |
| Terminal-Bench | 34.8% | — |
| Berkeley Function Calling Leaderboard | 55.5% | — |
| OSWorld 2.0 | — | 4.6% |
| ExploitBench | — | 18.4% |
| GBAEval | — | 0.9% |
| GDP.pdf | — | 12% |
Reasoning Kimi K2.6 leads
GPT-5 Mini: 23.9 (#168), Kimi K2.6: 40.5 (#55)
| Benchmark | GPT-5 Mini | Kimi K2.6 |
|---|---|---|
| CritPt | 0% | 8% |
| Chess Puzzles | 30% | 26% |
| LMArena Hard Prompts | 1380 | 1470 |
| Mystery Game Puzzles | 10% | 18% |
| DTBench | 80.5% | 90.9% |
| LMCA | 34.2% | 37.3% |
| Epoch Capabilities Index | 145.52 | 151.05 |
| ARC-AGI-2 | 4.4% | — |
| Kagi LLM Benchmark | 70.3% | — |
| NYT Connections (extended) | — | 87.2% |
| ARC-AGI-1 | 54.3% | — |
| EnigmaEval | 8.2% | — |
| EBR-Bench | — | 2.4% |
| ForecastBench | 61 | — |
Math Kimi K2.6 leads
GPT-5 Mini: 46.7 (#69), Kimi K2.6: 57.0 (#41)
| Benchmark | GPT-5 Mini | Kimi K2.6 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 46.7% | 57.2% |
| FrontierMath Tier 4 | 12.2% | 25.6% |
| OTIS Mock AIME 2024-2025 | 86.7% | 96.1% |
| ProofBench | 9% | 16% |
| LMArena Math | 1378 | 1475 |
| FrontierMath (Feb 2025 set) | 27.2% | 39% |
| FrontierMath Tier 4 (v1) | 6.3% | 14.6% |
| MathArena Final-Answer Competitions | — | 72.9% |
| Omni-MATH | 72.2% | — |
| MATH Level 5 | 97.8% | — |
Knowledge Kimi K2.6 leads
GPT-5 Mini: 45.6 (#86), Kimi K2.6: 54.0 (#54)
| Benchmark | GPT-5 Mini | Kimi K2.6 |
|---|---|---|
| GPQA Diamond | 75% | 90.8% |
| SimpleQA Verified | 21.6% | 34.9% |
| Vectara Hallucination Rate | 12.9% | 10.8% |
| LMArena Expert | 1379 | 1491 |
| Humanity's Last Exam | 19.4% | — |
| MMLU-Pro | 83.5% | — |
| Confabulations | 13.3% | — |
| GPQA (HELM) | 75.6% | — |
Multimodal GPT-5 Mini leads
GPT-5 Mini: 35.6 (#85), Kimi K2.6: 31.6 (#103)
| Benchmark | GPT-5 Mini | Kimi K2.6 |
|---|---|---|
| LMArena Vision | 1202 | 1283 |
| VPCT | 40.2% | — |
| Blueprint-Bench 2 | — | 3.9% |
| Furniture Assembly | — | 21.7% |
| LMArena Document | — | 1451 |
Multilingual Kimi K2.6 leads
GPT-5 Mini: 48.9 (#137), Kimi K2.6: 54.9 (#37)
| Benchmark | GPT-5 Mini | Kimi K2.6 |
|---|---|---|
| LMArena Non-English | 1363 | 1446 |
| LMArena Chinese | 1385 | 1521 |
| LMArena French | 1386 | 1471 |
| LMArena German | 1366 | 1450 |
| LMArena Japanese | 1341 | 1443 |
| LMArena Korean | 1308 | 1427 |
| LMArena Russian | 1362 | 1446 |
| LMArena Spanish | 1355 | 1464 |
Instruction Following Too close to call
GPT-5 Mini: 76.2 (#46), Kimi K2.6: 76.3 (#43)
| Benchmark | GPT-5 Mini | Kimi K2.6 |
|---|---|---|
| LMArena Instruction Following | 1357 | 1451 |
| IFEval | 92.7% | — |
Long Context Kimi K2.6 leads
GPT-5 Mini: 41.9 (#132), Kimi K2.6: 44.9 (#52)
| Benchmark | GPT-5 Mini | Kimi K2.6 |
|---|---|---|
| LMArena Longer Query | 1355 | 1468 |
| Fiction.LiveBench | 69.4% | — |
Writing & Preference Kimi K2.6 leads
GPT-5 Mini: 55.2 (#148), Kimi K2.6: 68.5 (#26)
| Benchmark | GPT-5 Mini | Kimi K2.6 |
|---|---|---|
| LMArena Text | 1373 | 1455 |
| LMArena Creative Writing | 1325 | 1434 |
| EQ-Bench Creative Writing | 1313 | 1725 |
| LMArena Multi-Turn | 1363 | 1453 |
| Short-Story Creative Writing | 83.1% | — |
| WildBench | 85.5% | — |
| EQ-Bench 4 | — | 1202 |
Frequently asked questions
Is GPT-5 Mini better than Kimi K2.6?
Kimi K2.6 is the stronger model overall, scoring 47.7 to 41.8 on the Noometry Index. GPT-5 Mini costs 2.5× less per token, which makes it the better buy when Kimi K2.6's lead doesn't matter for your workload.
Which is cheaper, GPT-5 Mini or Kimi K2.6?
GPT-5 Mini is cheaper. It lists at $0.25 per million input tokens and $2 per million output tokens; Kimi K2.6 lists at $0.95 and $4.
Is GPT-5 Mini or Kimi K2.6 better for coding?
Kimi K2.6 scores higher on coding benchmarks: 50.7 versus 40.1 in the Noometry coding category.
Which has the bigger context window?
GPT-5 Mini does, with 400K tokens against 262K.
How many benchmarks do GPT-5 Mini and Kimi K2.6 share?
39 benchmarks have published results for both models. GPT-5 Mini has 60 scored results on Noometry and Kimi K2.6 has 51.