Model comparison
Kimi K2.5 vs Kimi K2.6
Kimi K2.5 and Kimi K2.6 score almost the same on the Noometry Index (48.1 vs 47.7), so choose on price, context window or the category you care about most.
Last verified . 37 shared benchmarks.
Summary
- They share 37 benchmarks with published results for both. Kimi K2.5 scores higher in 3 categories and Kimi K2.6 in 7 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in agentic & tool use, where Kimi K2.5 leads 34.2 to 21.9.
- The biggest single-benchmark swing is NYT Connections (extended): 69.9% for Kimi K2.5 and 87.2% for Kimi K2.6.
- Kimi K2.5 is cheaper at $0.45 / $2.25 per million input/output tokens, against $0.95 / $4 for Kimi K2.6.
Side by side
| Kimi K2.5 | Kimi K2.6 | |
|---|---|---|
| Provider | Moonshot AI | Moonshot AI |
| Noometry Index | 48.1 | 47.7 |
| Released | 2026-01-27 | 2026-04-20 |
| Weights | Open | Open |
| Context window | 262K | 262K |
| Max output | 262K | 262K |
| Input $ / M tokens | $0.45 | $0.95 |
| Output $ / M tokens | $2.25 | $4 |
| Results tracked | 51 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K2.6 leads
Kimi K2.5: 48.8 (#53), Kimi K2.6: 50.7 (#43)
| Benchmark | Kimi K2.5 | Kimi K2.6 |
|---|---|---|
| SWE-bench Verified | 73.8% | 76.7% |
| LMArena WebDev | 1437 | 1509 |
| SciCode | 49% | 53.5% |
| WeirdML | 45.6% | 55.9% |
| LMArena Coding | 1474 | 1488 |
| ALE-Bench | 821.65 | 1,093 |
| SWE-bench Verified (bash only) | 70.8% | — |
| SWE-bench Multilingual | 67.3% | — |
Agentic & Tool Use Kimi K2.5 leads
Kimi K2.5: 34.2 (#48), Kimi K2.6: 21.9 (#137)
| Benchmark | Kimi K2.5 | Kimi K2.6 |
|---|---|---|
| Vending-Bench 2 | 1,198 | 6,205 |
| Terminal-Bench | 43.2% | — |
| OSWorld 2.0 | — | 4.6% |
| OSWorld | 63.3% | — |
| ExploitBench | — | 18.4% |
| GBAEval | — | 0.9% |
| GDP.pdf | — | 12% |
Reasoning Kimi K2.6 leads
Kimi K2.5: 31.2 (#80), Kimi K2.6: 40.5 (#55)
| Benchmark | Kimi K2.5 | Kimi K2.6 |
|---|---|---|
| NYT Connections (extended) | 69.9% | 87.2% |
| CritPt | 3.1% | 8% |
| Chess Puzzles | 12% | 26% |
| LMArena Hard Prompts | 1453 | 1470 |
| Epoch Capabilities Index | 148.03 | 151.05 |
| ARC-AGI-2 | 11.8% | — |
| SimpleBench | 46.8% | — |
| Kagi LLM Benchmark | 78.5% | — |
| ARC-AGI-1 | 65.3% | — |
| EnigmaEval | 3.4% | — |
| Thematic Generalization | 69.4% | — |
| EBR-Bench | — | 2.4% |
| Mystery Game Puzzles | — | 18% |
| DTBench | — | 90.9% |
| LMCA | — | 37.3% |
Math Kimi K2.6 leads
Kimi K2.5: 51.8 (#53), Kimi K2.6: 57.0 (#41)
| Benchmark | Kimi K2.5 | Kimi K2.6 |
|---|---|---|
| MathArena Final-Answer Competitions | 62.3% | 72.9% |
| OTIS Mock AIME 2024-2025 | 92.2% | 96.1% |
| LMArena Math | 1470 | 1475 |
| FrontierMath (Feb 2025 set) | 27.9% | 39% |
| FrontierMath Tier 4 (v1) | 4.2% | 14.6% |
| FrontierMath (Tiers 1-3) | — | 57.2% |
| FrontierMath Tier 4 | — | 25.6% |
| ProofBench | — | 16% |
Knowledge Too close to call
Kimi K2.5: 53.6 (#56), Kimi K2.6: 54.0 (#54)
| Benchmark | Kimi K2.5 | Kimi K2.6 |
|---|---|---|
| GPQA Diamond | 87.6% | 90.8% |
| SimpleQA Verified | 34.3% | 34.9% |
| Vectara Hallucination Rate | 14.2% | 10.8% |
| LMArena Expert | 1466 | 1491 |
| Humanity's Last Exam | 24.4% | — |
Multimodal Kimi K2.5 leads
Kimi K2.5: 41.1 (#39), Kimi K2.6: 31.6 (#103)
| Benchmark | Kimi K2.5 | Kimi K2.6 |
|---|---|---|
| LMArena Vision | 1269 | 1283 |
| LMArena Document | 1430 | 1451 |
| Blueprint-Bench 2 | — | 3.9% |
| Furniture Assembly | — | 21.7% |
Multilingual Too close to call
Kimi K2.5: 53.9 (#53), Kimi K2.6: 54.9 (#37)
| Benchmark | Kimi K2.5 | Kimi K2.6 |
|---|---|---|
| LMArena Non-English | 1433 | 1446 |
| LMArena Chinese | 1495 | 1521 |
| LMArena French | 1454 | 1471 |
| LMArena German | 1441 | 1450 |
| LMArena Japanese | 1421 | 1443 |
| LMArena Korean | 1410 | 1427 |
| LMArena Russian | 1435 | 1446 |
| LMArena Spanish | 1450 | 1464 |
Instruction Following Too close to call
Kimi K2.5: 75.3 (#64), Kimi K2.6: 76.3 (#43)
| Benchmark | Kimi K2.5 | Kimi K2.6 |
|---|---|---|
| LMArena Instruction Following | 1431 | 1451 |
Long Context Kimi K2.5 leads
Kimi K2.5: 52.1 (#7), Kimi K2.6: 44.9 (#52)
| Benchmark | Kimi K2.5 | Kimi K2.6 |
|---|---|---|
| LMArena Longer Query | 1445 | 1468 |
| Fiction.LiveBench | 86.1% | — |
| CL-bench | 19.3% | — |
| CL-bench Life | 13.2% | — |
Writing & Preference Kimi K2.6 leads
Kimi K2.5: 65.1 (#53), Kimi K2.6: 68.5 (#26)
| Benchmark | Kimi K2.5 | Kimi K2.6 |
|---|---|---|
| LMArena Text | 1445 | 1455 |
| LMArena Creative Writing | 1423 | 1434 |
| EQ-Bench Creative Writing | 1579 | 1725 |
| LMArena Multi-Turn | 1444 | 1453 |
| EQ-Bench 4 | — | 1202 |
Frequently asked questions
Is Kimi K2.5 better than Kimi K2.6?
Kimi K2.5 and Kimi K2.6 score almost the same on the Noometry Index (48.1 vs 47.7), so choose on price, context window or the category you care about most.
Which is cheaper, Kimi K2.5 or Kimi K2.6?
Kimi K2.5 is cheaper. It lists at $0.45 per million input tokens and $2.25 per million output tokens; Kimi K2.6 lists at $0.95 and $4.
Is Kimi K2.5 or Kimi K2.6 better for coding?
Kimi K2.6 scores higher on coding benchmarks: 50.7 versus 48.8 in the Noometry coding category.
Which has the bigger context window?
Both accept 262K tokens.
How many benchmarks do Kimi K2.5 and Kimi K2.6 share?
37 benchmarks have published results for both models. Kimi K2.5 has 51 scored results on Noometry and Kimi K2.6 has 51.