Model comparison
DeepSeek-R1 vs Kimi K2 Thinking Turbo
Kimi K2 Thinking Turbo is the stronger model overall, scoring 45.4 to 42.0 on the Noometry Index.
Last verified . 19 shared benchmarks.
DeepSeekSummary
- They share 19 benchmarks with published results for both. DeepSeek-R1 scores higher in 4 categories and Kimi K2 Thinking Turbo in 4 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Kimi K2 Thinking Turbo leads 31.4 to 18.7.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 66.4% for DeepSeek-R1 and 83.1% for Kimi K2 Thinking Turbo.
- Kimi K2 Thinking Turbo has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-R1 | Kimi K2 Thinking Turbo | |
|---|---|---|
| Provider | DeepSeek | Moonshot AI |
| Noometry Index | 42.0 | 45.4 |
| Released | 2025-01-20 | 2025-11-06 |
| Weights | Proprietary | Open |
| Context window | 164K | — |
| Max output | 64K | — |
| Input $ / M tokens | $0.50 | — |
| Output $ / M tokens | $2.15 | — |
| Results tracked | 52 | 21 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-R1 leads
DeepSeek-R1: 45.9 (#63), Kimi K2 Thinking Turbo: 38.1 (#166)
| Benchmark | DeepSeek-R1 | Kimi K2 Thinking Turbo |
|---|---|---|
| LMArena Coding | 1427 | 1454 |
| Aider Polyglot | 71.4% | — |
| LMArena WebDev | — | 1322 |
| SciCode | 35.7% | — |
| WeirdML | 41.6% | — |
| LiveBench Coding | 66.7% | — |
| ALE-Bench | 804.12 | — |
| AlgoTune | 1.7 | — |
Agentic & Tool Use Not comparable
DeepSeek-R1: 30.4 (#68), Kimi K2 Thinking Turbo: —
| Benchmark | DeepSeek-R1 | Kimi K2 Thinking Turbo |
|---|---|---|
| DeepResearch Bench | 35.1% | — |
| BALROG | 34.9% | — |
| METR Time Horizons | 53.8% | — |
Reasoning Kimi K2 Thinking Turbo leads
DeepSeek-R1: 18.7 (#253), Kimi K2 Thinking Turbo: 31.4 (#72)
| Benchmark | DeepSeek-R1 | Kimi K2 Thinking Turbo |
|---|---|---|
| LMArena Hard Prompts | 1415 | 1428 |
| ARC-AGI-2 | 1.3% | — |
| SimpleBench | 40.8% | — |
| Kagi LLM Benchmark | 69.4% | — |
| ARC-AGI-1 | 21.2% | — |
| CritPt | 1.1% | — |
| Chess Puzzles | — | 20% |
| LiveBench Reasoning | 83.2% | — |
| LiveBench Data Analysis | 69.8% | — |
| Epoch Capabilities Index | 141.3 | — |
| ForecastBench | 60 | — |
| LiveBench | 71.6% | — |
Math Kimi K2 Thinking Turbo leads
DeepSeek-R1: 42.0 (#70), Kimi K2 Thinking Turbo: 45.5 (#62)
| Benchmark | DeepSeek-R1 | Kimi K2 Thinking Turbo |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.4% | 83.1% |
| LMArena Math | 1401 | 1429 |
| Omni-MATH | 42.4% | — |
| LiveBench Math | 80.7% | — |
| MATH Level 5 | 96.6% | — |
Knowledge Kimi K2 Thinking Turbo leads
DeepSeek-R1: 44.3 (#81), Kimi K2 Thinking Turbo: 50.9 (#64)
| Benchmark | DeepSeek-R1 | Kimi K2 Thinking Turbo |
|---|---|---|
| GPQA Diamond | 76.3% | 84.2% |
| LMArena Expert | 1394 | 1439 |
| MMLU-Pro | 79.3% | — |
| Confabulations | 12.7% | — |
| Vectara Hallucination Rate | 11.3% | — |
| GPQA (HELM) | 66.6% | — |
Multilingual DeepSeek-R1 leads
DeepSeek-R1: 52.6 (#80), Kimi K2 Thinking Turbo: 51.6 (#104)
| Benchmark | DeepSeek-R1 | Kimi K2 Thinking Turbo |
|---|---|---|
| LMArena Non-English | 1412 | 1398 |
| LMArena Chinese | 1443 | 1457 |
| LMArena French | 1415 | 1425 |
| LMArena German | 1405 | 1390 |
| LMArena Japanese | 1394 | 1356 |
| LMArena Korean | 1359 | 1329 |
| LMArena Russian | 1423 | 1390 |
| LMArena Spanish | 1410 | 1407 |
Instruction Following Kimi K2 Thinking Turbo leads
DeepSeek-R1: 72.2 (#138), Kimi K2 Thinking Turbo: 74.2 (#103)
| Benchmark | DeepSeek-R1 | Kimi K2 Thinking Turbo |
|---|---|---|
| LMArena Instruction Following | 1382 | 1403 |
| LiveBench Instruction Following | 80.5% | — |
| IFEval | 78.4% | — |
Long Context DeepSeek-R1 leads
DeepSeek-R1: 45.4 (#36), Kimi K2 Thinking Turbo: 43.2 (#99)
| Benchmark | DeepSeek-R1 | Kimi K2 Thinking Turbo |
|---|---|---|
| LMArena Longer Query | 1391 | 1415 |
| Fiction.LiveBench | 75% | — |
Writing & Preference DeepSeek-R1 leads
DeepSeek-R1: 61.5 (#84), Kimi K2 Thinking Turbo: 60.2 (#99)
| Benchmark | DeepSeek-R1 | Kimi K2 Thinking Turbo |
|---|---|---|
| LMArena Text | 1428 | 1415 |
| LMArena Creative Writing | 1406 | 1373 |
| LMArena Multi-Turn | 1404 | 1414 |
| Short-Story Creative Writing | 83% | — |
| EQ-Bench Creative Writing | 1500 | — |
| WildBench | 82.8% | — |
| LiveBench Language | 48.5% | — |
Frequently asked questions
Is DeepSeek-R1 better than Kimi K2 Thinking Turbo?
Kimi K2 Thinking Turbo is the stronger model overall, scoring 45.4 to 42.0 on the Noometry Index.
Is DeepSeek-R1 or Kimi K2 Thinking Turbo better for coding?
DeepSeek-R1 scores higher on coding benchmarks: 45.9 versus 38.1 in the Noometry coding category.
How many benchmarks do DeepSeek-R1 and Kimi K2 Thinking Turbo share?
19 benchmarks have published results for both models. DeepSeek-R1 has 52 scored results on Noometry and Kimi K2 Thinking Turbo has 21.