Model comparison
Claude Opus 5 vs Kimi K2.6
Claude Opus 5 is the stronger model overall, scoring 67.8 to 47.7 on the Noometry Index. Kimi K2.6 costs 5.8× less per token, which makes it the better buy when Claude Opus 5's lead doesn't matter for your workload.
Last verified . 45 shared benchmarks.
Summary
- They share 45 benchmarks with published results for both. Claude Opus 5 scores higher in 10 categories and Kimi K2.6 in 0 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Claude Opus 5 leads 77.2 to 40.5.
- The biggest single-benchmark swing is ProofBench: 99% for Claude Opus 5 and 16% for Kimi K2.6.
- Kimi K2.6 is cheaper at $0.95 / $4 per million input/output tokens, against $5 / $25 for Claude Opus 5.
- Claude Opus 5 accepts more context: 1M tokens versus 262K.
- Kimi K2.6 has downloadable open weights; the other is API-only.
Side by side
| Claude Opus 5 | Kimi K2.6 | |
|---|---|---|
| Provider | Anthropic | Moonshot AI |
| Noometry Index | 67.8 | 47.7 |
| Released | 2026-07-24 | 2026-04-20 |
| Weights | Proprietary | Open |
| Context window | 1M | 262K |
| Max output | 128K | 262K |
| Input $ / M tokens | $5 | $0.95 |
| Output $ / M tokens | $25 | $4 |
| Results tracked | 57 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Opus 5 leads
Claude Opus 5: 67.5 (#5), Kimi K2.6: 50.7 (#43)
| Benchmark | Claude Opus 5 | Kimi K2.6 |
|---|---|---|
| LMArena WebDev | 1691 | 1509 |
| SciCode | 56.4% | 53.5% |
| WeirdML | 91.8% | 55.9% |
| LMArena Coding | 1534 | 1488 |
| ALE-Bench | 2,165 | 1,093 |
| SWE-bench Verified | — | 76.7% |
| DeepSWE | 73.6% | — |
| FrontierCode | 53.4% | — |
| CursorBench | 46.6% | — |
| FrontierSWE | 52% | — |
Agentic & Tool Use Claude Opus 5 leads
Claude Opus 5: 55.6 (#1), Kimi K2.6: 21.9 (#137)
| Benchmark | Claude Opus 5 | Kimi K2.6 |
|---|---|---|
| OSWorld 2.0 | 31.4% | 4.6% |
| GBAEval | 79.6% | 0.9% |
| GDP.pdf | 24% | 12% |
| Vending-Bench 2 | 11,182 | 6,205 |
| APEX-Agents | 65.8% | — |
| τ²-bench Banking | 48.7% | — |
| PostTrainBench | 35% | — |
| BALROG | 63.4% | — |
| ExploitBench | — | 18.4% |
Reasoning Claude Opus 5 leads
Claude Opus 5: 77.2 (#4), Kimi K2.6: 40.5 (#55)
| Benchmark | Claude Opus 5 | Kimi K2.6 |
|---|---|---|
| NYT Connections (extended) | 94.3% | 87.2% |
| CritPt | 29.1% | 8% |
| Chess Puzzles | 42% | 26% |
| EBR-Bench | 45.7% | 2.4% |
| LMArena Hard Prompts | 1526 | 1470 |
| Mystery Game Puzzles | 59% | 18% |
| DTBench | 97.9% | 90.9% |
| LMCA | 64.5% | 37.3% |
| Epoch Capabilities Index | 162.78 | 151.05 |
| ARC-AGI-2 | 90.4% | — |
| SimpleBench | 80.6% | — |
| ARC-AGI-1 | 97.5% | — |
| Bench to the Future 3 | 0.12 | — |
Math Claude Opus 5 leads
Claude Opus 5: 86.2 (#8), Kimi K2.6: 57.0 (#41)
| Benchmark | Claude Opus 5 | Kimi K2.6 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 85.6% | 57.2% |
| FrontierMath Tier 4 | 73.2% | 25.6% |
| OTIS Mock AIME 2024-2025 | 98.9% | 96.1% |
| ProofBench | 99% | 16% |
| LMArena Math | 1531 | 1475 |
| MathArena Final-Answer Competitions | — | 72.9% |
| FrontierMath (Feb 2025 set) | — | 39% |
| FrontierMath Tier 4 (v1) | — | 14.6% |
Knowledge Claude Opus 5 leads
Claude Opus 5: 66.8 (#9), Kimi K2.6: 54.0 (#54)
| Benchmark | Claude Opus 5 | Kimi K2.6 |
|---|---|---|
| GPQA Diamond | 93.9% | 90.8% |
| SimpleQA Verified | 59.9% | 34.9% |
| LMArena Expert | 1557 | 1491 |
| Vectara Hallucination Rate | — | 10.8% |
Multimodal Claude Opus 5 leads
Claude Opus 5: 50.8 (#8), Kimi K2.6: 31.6 (#103)
| Benchmark | Claude Opus 5 | Kimi K2.6 |
|---|---|---|
| LMArena Vision | 1319 | 1283 |
| Blueprint-Bench 2 | 30.4% | 3.9% |
| Furniture Assembly | 60.8% | 21.7% |
| LMArena Document | 1516 | 1451 |
Multilingual Claude Opus 5 leads
Claude Opus 5: 58.8 (#4), Kimi K2.6: 54.9 (#37)
| Benchmark | Claude Opus 5 | Kimi K2.6 |
|---|---|---|
| LMArena Non-English | 1501 | 1446 |
| LMArena Chinese | 1574 | 1521 |
| LMArena French | 1519 | 1471 |
| LMArena German | 1524 | 1450 |
| LMArena Japanese | 1516 | 1443 |
| LMArena Korean | 1521 | 1427 |
| LMArena Russian | 1507 | 1446 |
| LMArena Spanish | 1519 | 1464 |
Instruction Following Claude Opus 5 leads
Claude Opus 5: 79.2 (#7), Kimi K2.6: 76.3 (#43)
| Benchmark | Claude Opus 5 | Kimi K2.6 |
|---|---|---|
| LMArena Instruction Following | 1517 | 1451 |
Long Context Claude Opus 5 leads
Claude Opus 5: 46.5 (#21), Kimi K2.6: 44.9 (#52)
| Benchmark | Claude Opus 5 | Kimi K2.6 |
|---|---|---|
| LMArena Longer Query | 1515 | 1468 |
Writing & Preference Claude Opus 5 leads
Claude Opus 5: 79.2 (#1), Kimi K2.6: 68.5 (#26)
| Benchmark | Claude Opus 5 | Kimi K2.6 |
|---|---|---|
| LMArena Text | 1507 | 1455 |
| LMArena Creative Writing | 1491 | 1434 |
| EQ-Bench Creative Writing | 2133 | 1725 |
| EQ-Bench 4 | 1385 | 1202 |
| LMArena Multi-Turn | 1499 | 1453 |
Frequently asked questions
Is Claude Opus 5 better than Kimi K2.6?
Claude Opus 5 is the stronger model overall, scoring 67.8 to 47.7 on the Noometry Index. Kimi K2.6 costs 5.8× less per token, which makes it the better buy when Claude Opus 5's lead doesn't matter for your workload.
Which is cheaper, Claude Opus 5 or Kimi K2.6?
Kimi K2.6 is cheaper. It lists at $0.95 per million input tokens and $4 per million output tokens; Claude Opus 5 lists at $5 and $25.
Is Claude Opus 5 or Kimi K2.6 better for coding?
Claude Opus 5 scores higher on coding benchmarks: 67.5 versus 50.7 in the Noometry coding category.
Which has the bigger context window?
Claude Opus 5 does, with 1M tokens against 262K.
How many benchmarks do Claude Opus 5 and Kimi K2.6 share?
45 benchmarks have published results for both models. Claude Opus 5 has 57 scored results on Noometry and Kimi K2.6 has 51.