Model comparison
Claude Opus 5 vs Kimi K3
Claude Opus 5 is the stronger model overall, scoring 67.8 to 59.5 on the Noometry Index. Kimi K3 costs 1.7× less per token, which makes it the better buy when Claude Opus 5's lead doesn't matter for your workload.
Last verified . 50 shared benchmarks.
Summary
- They share 50 benchmarks with published results for both. Claude Opus 5 scores higher in 10 categories and Kimi K3 in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Claude Opus 5 leads 77.2 to 63.0.
- The biggest single-benchmark swing is FrontierMath Tier 4: 73.2% for Claude Opus 5 and 39% for Kimi K3.
- Kimi K3 is cheaper at $3 / $15 per million input/output tokens, against $5 / $25 for Claude Opus 5.
- Kimi K3 accepts more context: 1.05M tokens versus 1M.
- Kimi K3 has downloadable open weights; the other is API-only.
Side by side
| Claude Opus 5 | Kimi K3 | |
|---|---|---|
| Provider | Anthropic | Moonshot AI |
| Noometry Index | 67.8 | 59.5 |
| Released | 2026-07-24 | 2026-07-16 |
| Weights | Proprietary | Open |
| Context window | 1M | 1.05M |
| Max output | 128K | 1.05M |
| Input $ / M tokens | $5 | $3 |
| Output $ / M tokens | $25 | $15 |
| Results tracked | 57 | 53 |
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 K3: 61.0 (#10)
| Benchmark | Claude Opus 5 | Kimi K3 |
|---|---|---|
| DeepSWE | 73.6% | 68.5% |
| FrontierCode | 53.4% | 44.2% |
| LMArena WebDev | 1691 | 1654 |
| FrontierSWE | 52% | 25.9% |
| SciCode | 56.4% | 59.5% |
| WeirdML | 91.8% | 82.6% |
| LMArena Coding | 1534 | 1508 |
| ALE-Bench | 2,165 | 1,524 |
| CursorBench | 46.6% | — |
Agentic & Tool Use Claude Opus 5 leads
Claude Opus 5: 55.6 (#1), Kimi K3: 41.8 (#20)
| Benchmark | Claude Opus 5 | Kimi K3 |
|---|---|---|
| APEX-Agents | 65.8% | 50.6% |
| τ²-bench Banking | 48.7% | 37.1% |
| PostTrainBench | 35% | 32% |
| GBAEval | 79.6% | 48.3% |
| GDP.pdf | 24% | 19% |
| Vending-Bench 2 | 11,182 | 5,165 |
| OSWorld 2.0 | 31.4% | — |
| BALROG | 63.4% | — |
Reasoning Claude Opus 5 leads
Claude Opus 5: 77.2 (#4), Kimi K3: 63.0 (#17)
| Benchmark | Claude Opus 5 | Kimi K3 |
|---|---|---|
| ARC-AGI-2 | 90.4% | 60.4% |
| SimpleBench | 80.6% | 60.7% |
| NYT Connections (extended) | 94.3% | 93.6% |
| ARC-AGI-1 | 97.5% | 94.5% |
| CritPt | 29.1% | 23.4% |
| Chess Puzzles | 42% | 39% |
| LMArena Hard Prompts | 1526 | 1496 |
| Mystery Game Puzzles | 59% | 26% |
| DTBench | 97.9% | 91.2% |
| LMCA | 64.5% | 52.7% |
| Epoch Capabilities Index | 162.78 | 157.45 |
| EBR-Bench | 45.7% | — |
| Surface Evolver Bench | — | 95% |
| Bench to the Future 3 | 0.12 | — |
| ForecastBench | — | 61.1 |
Math Claude Opus 5 leads
Claude Opus 5: 86.2 (#8), Kimi K3: 74.2 (#16)
| Benchmark | Claude Opus 5 | Kimi K3 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 85.6% | 72.2% |
| FrontierMath Tier 4 | 73.2% | 39% |
| OTIS Mock AIME 2024-2025 | 98.9% | 97.2% |
| ProofBench | 99% | 87% |
| LMArena Math | 1531 | 1491 |
| MathArena Final-Answer Competitions | — | 87.8% |
Knowledge Claude Opus 5 leads
Claude Opus 5: 66.8 (#9), Kimi K3: 63.2 (#21)
| Benchmark | Claude Opus 5 | Kimi K3 |
|---|---|---|
| GPQA Diamond | 93.9% | 93.1% |
| SimpleQA Verified | 59.9% | 50.6% |
| LMArena Expert | 1557 | 1521 |
Multimodal Claude Opus 5 leads
Claude Opus 5: 50.8 (#8), Kimi K3: 37.8 (#70)
| Benchmark | Claude Opus 5 | Kimi K3 |
|---|---|---|
| Blueprint-Bench 2 | 30.4% | 29.5% |
| Furniture Assembly | 60.8% | 34.2% |
| LMArena Vision | 1319 | — |
| LMArena Document | 1516 | — |
Multilingual Claude Opus 5 leads
Claude Opus 5: 58.8 (#4), Kimi K3: 56.3 (#21)
| Benchmark | Claude Opus 5 | Kimi K3 |
|---|---|---|
| LMArena Non-English | 1501 | 1466 |
| LMArena Chinese | 1574 | 1529 |
| LMArena French | 1519 | 1491 |
| LMArena German | 1524 | 1488 |
| LMArena Japanese | 1516 | 1487 |
| LMArena Korean | 1521 | 1458 |
| LMArena Russian | 1507 | 1482 |
| LMArena Spanish | 1519 | 1472 |
Instruction Following Claude Opus 5 leads
Claude Opus 5: 79.2 (#7), Kimi K3: 77.7 (#14)
| Benchmark | Claude Opus 5 | Kimi K3 |
|---|---|---|
| LMArena Instruction Following | 1517 | 1483 |
Long Context Too close to call
Claude Opus 5: 46.5 (#21), Kimi K3: 45.8 (#29)
| Benchmark | Claude Opus 5 | Kimi K3 |
|---|---|---|
| LMArena Longer Query | 1515 | 1494 |
Writing & Preference Claude Opus 5 leads
Claude Opus 5: 79.2 (#1), Kimi K3: 76.6 (#4)
| Benchmark | Claude Opus 5 | Kimi K3 |
|---|---|---|
| LMArena Text | 1507 | 1476 |
| LMArena Creative Writing | 1491 | 1454 |
| EQ-Bench Creative Writing | 2133 | 2082 |
| EQ-Bench 4 | 1385 | 1339 |
| LMArena Multi-Turn | 1499 | 1488 |
Frequently asked questions
Is Claude Opus 5 better than Kimi K3?
Claude Opus 5 is the stronger model overall, scoring 67.8 to 59.5 on the Noometry Index. Kimi K3 costs 1.7× 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 K3?
Kimi K3 is cheaper. It lists at $3 per million input tokens and $15 per million output tokens; Claude Opus 5 lists at $5 and $25.
Is Claude Opus 5 or Kimi K3 better for coding?
Claude Opus 5 scores higher on coding benchmarks: 67.5 versus 61.0 in the Noometry coding category.
Which has the bigger context window?
Kimi K3 does, with 1.05M tokens against 1M.
How many benchmarks do Claude Opus 5 and Kimi K3 share?
50 benchmarks have published results for both models. Claude Opus 5 has 57 scored results on Noometry and Kimi K3 has 53.