Model comparison
DeepSeek-V3.2-Exp vs Kimi K2.6
Kimi K2.6 is the stronger model overall, scoring 47.5 to 44.9 on the Noometry Index. DeepSeek-V3.2-Exp costs 5.9× less per token, which makes it the better buy when Kimi K2.6's lead doesn't matter for your workload.
Last verified . 35 shared benchmarks.
DeepSeekSummary
- They share 35 benchmarks with published results for both. DeepSeek-V3.2-Exp scores higher in 2 categories and Kimi K2.6 in 7 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Kimi K2.6 leads 39.6 to 22.0.
- The biggest single-benchmark swing is NYT Connections (extended): 36.7% for DeepSeek-V3.2-Exp and 87.2% for Kimi K2.6.
- DeepSeek-V3.2-Exp is cheaper at $0.26 / $0.38 per million input/output tokens, against $0.95 / $4 for Kimi K2.6.
- Kimi K2.6 accepts more context: 262K tokens versus 164K.
Side by side
| DeepSeek-V3.2-Exp | Kimi K2.6 | |
|---|---|---|
| Provider | DeepSeek | Moonshot AI |
| Noometry Index | 44.9 | 47.5 |
| Released | 2025-09-29 | 2026-04-20 |
| Weights | Open | Open |
| Context window | 164K | 262K |
| Max output | 66K | 262K |
| Input $ / M tokens | $0.26 | $0.95 |
| Output $ / M tokens | $0.38 | $4 |
| Results tracked | 48 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K2.6 leads
DeepSeek-V3.2-Exp: 46.4 (#61), Kimi K2.6: 50.5 (#41)
| Benchmark | DeepSeek-V3.2-Exp | Kimi K2.6 |
|---|---|---|
| LMArena WebDev | 1362 | 1509 |
| SciCode | 38.9% | 53.5% |
| WeirdML | 39.5% | 55.9% |
| LMArena Coding | 1454 | 1488 |
| SWE-bench Verified | — | 76.7% |
| SWE-bench Verified (bash only) | 70% | — |
| Aider Polyglot | 74.2% | — |
| SWE-bench Multilingual | 59% | — |
| ALE-Bench | — | 1,093 |
Agentic & Tool Use DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 37.9 (#31), Kimi K2.6: 21.8 (#128)
| Benchmark | DeepSeek-V3.2-Exp | Kimi K2.6 |
|---|---|---|
| Vending-Bench 2 | 1,034 | 6,205 |
| Terminal-Bench | 39.6% | — |
| Berkeley Function Calling Leaderboard | 56.7% | — |
| OSWorld 2.0 | — | 4.6% |
| TheAgentCompany | 42.9% | — |
| ExploitBench | — | 18.4% |
| GBAEval | — | 0.9% |
| GDP.pdf | — | 12% |
Reasoning Kimi K2.6 leads
DeepSeek-V3.2-Exp: 22.0 (#185), Kimi K2.6: 39.6 (#51)
| Benchmark | DeepSeek-V3.2-Exp | Kimi K2.6 |
|---|---|---|
| NYT Connections (extended) | 36.7% | 87.2% |
| CritPt | 2.9% | 8% |
| Chess Puzzles | 14% | 26% |
| LMArena Hard Prompts | 1433 | 1470 |
| DTBench | 87.7% | 90.9% |
| LMCA | 29.1% | 37.3% |
| Epoch Capabilities Index | 146.28 | 151.03 |
| ARC-AGI-2 | 4% | — |
| Kagi LLM Benchmark | 52.2% | — |
| ARC-AGI-1 | 57% | — |
| Thematic Generalization | 65% | — |
| EBR-Bench | — | 2.4% |
| Mystery Game Puzzles | — | 18% |
Math Kimi K2.6 leads
DeepSeek-V3.2-Exp: 40.8 (#75), Kimi K2.6: 56.8 (#34)
| Benchmark | DeepSeek-V3.2-Exp | Kimi K2.6 |
|---|---|---|
| MathArena Final-Answer Competitions | 57.7% | 72.9% |
| OTIS Mock AIME 2024-2025 | 87.8% | 96.1% |
| ProofBench | 8% | 16% |
| LMArena Math | 1435 | 1475 |
| FrontierMath (Feb 2025 set) | 22.1% | 39% |
| FrontierMath Tier 4 (v1) | 2.1% | 14.6% |
| FrontierMath (Tiers 1-3) | — | 57.2% |
| FrontierMath Tier 4 | — | 25.6% |
Knowledge Kimi K2.6 leads
DeepSeek-V3.2-Exp: 51.6 (#61), Kimi K2.6: 53.9 (#48)
| Benchmark | DeepSeek-V3.2-Exp | Kimi K2.6 |
|---|---|---|
| GPQA Diamond | 83.4% | 90.8% |
| Vectara Hallucination Rate | 5.3% | 10.8% |
| LMArena Expert | 1436 | 1491 |
| SimpleQA Verified | — | 34.9% |
Multimodal Not comparable
DeepSeek-V3.2-Exp: —, Kimi K2.6: 31.8 (#95)
| Benchmark | DeepSeek-V3.2-Exp | Kimi K2.6 |
|---|---|---|
| LMArena Vision | — | 1282 |
| Blueprint-Bench 2 | — | 3.9% |
| Furniture Assembly | — | 21.7% |
| LMArena Document | — | 1451 |
Multilingual Kimi K2.6 leads
DeepSeek-V3.2-Exp: 52.4 (#85), Kimi K2.6: 55.1 (#35)
| Benchmark | DeepSeek-V3.2-Exp | Kimi K2.6 |
|---|---|---|
| LMArena Non-English | 1409 | 1446 |
| LMArena Chinese | 1461 | 1521 |
| LMArena French | 1435 | 1471 |
| LMArena German | 1440 | 1449 |
| LMArena Japanese | 1377 | 1442 |
| LMArena Korean | 1372 | 1428 |
| LMArena Russian | 1423 | 1446 |
| LMArena Spanish | 1441 | 1463 |
Instruction Following Kimi K2.6 leads
DeepSeek-V3.2-Exp: 74.7 (#89), Kimi K2.6: 76.5 (#40)
| Benchmark | DeepSeek-V3.2-Exp | Kimi K2.6 |
|---|---|---|
| LMArena Instruction Following | 1413 | 1451 |
Long Context DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 47.6 (#15), Kimi K2.6: 44.9 (#50)
| Benchmark | DeepSeek-V3.2-Exp | Kimi K2.6 |
|---|---|---|
| LMArena Longer Query | 1428 | 1468 |
| Fiction.LiveBench | 83.3% | — |
| CL-bench | 13.2% | — |
| CL-bench Life | 9.5% | — |
Writing & Preference Kimi K2.6 leads
DeepSeek-V3.2-Exp: 62.5 (#73), Kimi K2.6: 68.7 (#25)
| Benchmark | DeepSeek-V3.2-Exp | Kimi K2.6 |
|---|---|---|
| LMArena Text | 1425 | 1455 |
| LMArena Creative Writing | 1403 | 1434 |
| EQ-Bench Creative Writing | 1515 | 1725 |
| LMArena Multi-Turn | 1427 | 1453 |
| EQ-Bench 4 | — | 1202 |
Frequently asked questions
Is DeepSeek-V3.2-Exp better than Kimi K2.6?
Kimi K2.6 is the stronger model overall, scoring 47.5 to 44.9 on the Noometry Index. DeepSeek-V3.2-Exp costs 5.9× less per token, which makes it the better buy when Kimi K2.6's lead doesn't matter for your workload.
Which is cheaper, DeepSeek-V3.2-Exp or Kimi K2.6?
DeepSeek-V3.2-Exp is cheaper. It lists at $0.26 per million input tokens and $0.38 per million output tokens; Kimi K2.6 lists at $0.95 and $4.
Is DeepSeek-V3.2-Exp or Kimi K2.6 better for coding?
Kimi K2.6 scores higher on coding benchmarks: 50.5 versus 46.4 in the Noometry coding category.
Which has the bigger context window?
Kimi K2.6 does, with 262K tokens against 164K.
How many benchmarks do DeepSeek-V3.2-Exp and Kimi K2.6 share?
35 benchmarks have published results for both models. DeepSeek-V3.2-Exp has 48 scored results on Noometry and Kimi K2.6 has 51.