Model comparison
Grok 4.20 Multi-Agent vs Kimi K2.6
Kimi K2.6 is the stronger model overall, scoring 47.5 to 45.5 on the Noometry Index.
Last verified . 19 shared benchmarks.
Summary
- They share 19 benchmarks with published results for both. Grok 4.20 Multi-Agent scores higher in 2 categories and Kimi K2.6 in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Kimi K2.6 leads 56.8 to 36.6.
- Grok 4.20 Multi-Agent is cheaper at $1.25 / $2.50 per million input/output tokens, against $0.95 / $4 for Kimi K2.6.
- Grok 4.20 Multi-Agent accepts more context: 1M tokens versus 262K.
- Kimi K2.6 has downloadable open weights; the other is API-only.
Side by side
| Grok 4.20 Multi-Agent | Kimi K2.6 | |
|---|---|---|
| Provider | xAI | Moonshot AI |
| Noometry Index | 45.5 | 47.5 |
| Released | 2026-03-09 | 2026-04-20 |
| Weights | Proprietary | Open |
| Context window | 1M | 262K |
| Max output | 30K | 262K |
| Input $ / M tokens | $1.25 | $0.95 |
| Output $ / M tokens | $2.50 | $4 |
| Results tracked | 20 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K2.6 leads
Grok 4.20 Multi-Agent: 42.4 (#91), Kimi K2.6: 50.5 (#41)
| Benchmark | Grok 4.20 Multi-Agent | Kimi K2.6 |
|---|---|---|
| LMArena Coding | 1458 | 1488 |
| SWE-bench Verified | — | 76.7% |
| LMArena WebDev | — | 1509 |
| SciCode | — | 53.5% |
| WeirdML | — | 55.9% |
| ALE-Bench | — | 1,093 |
Agentic & Tool Use Not comparable
Grok 4.20 Multi-Agent: —, Kimi K2.6: 21.8 (#128)
| Benchmark | Grok 4.20 Multi-Agent | Kimi K2.6 |
|---|---|---|
| OSWorld 2.0 | — | 4.6% |
| ExploitBench | — | 18.4% |
| GBAEval | — | 0.9% |
| GDP.pdf | — | 12% |
| LMArena Search | 1204 | — |
| Vending-Bench 2 | — | 6,205 |
Reasoning Grok 4.20 Multi-Agent leads
Grok 4.20 Multi-Agent: 42.4 (#44), Kimi K2.6: 39.6 (#51)
| Benchmark | Grok 4.20 Multi-Agent | Kimi K2.6 |
|---|---|---|
| NYT Connections (extended) | 89.6% | 87.2% |
| LMArena Hard Prompts | 1448 | 1470 |
| CritPt | — | 8% |
| Chess Puzzles | — | 26% |
| EBR-Bench | — | 2.4% |
| Mystery Game Puzzles | — | 18% |
| DTBench | — | 90.9% |
| LMCA | — | 37.3% |
| Epoch Capabilities Index | — | 151.03 |
Math Kimi K2.6 leads
Grok 4.20 Multi-Agent: 36.6 (#106), Kimi K2.6: 56.8 (#34)
| Benchmark | Grok 4.20 Multi-Agent | Kimi K2.6 |
|---|---|---|
| LMArena Math | 1441 | 1475 |
| FrontierMath (Tiers 1-3) | — | 57.2% |
| FrontierMath Tier 4 | — | 25.6% |
| MathArena Final-Answer Competitions | — | 72.9% |
| OTIS Mock AIME 2024-2025 | — | 96.1% |
| ProofBench | — | 16% |
| FrontierMath (Feb 2025 set) | — | 39% |
| FrontierMath Tier 4 (v1) | — | 14.6% |
Knowledge Kimi K2.6 leads
Grok 4.20 Multi-Agent: 39.7 (#116), Kimi K2.6: 53.9 (#48)
| Benchmark | Grok 4.20 Multi-Agent | Kimi K2.6 |
|---|---|---|
| LMArena Expert | 1445 | 1491 |
| GPQA Diamond | — | 90.8% |
| SimpleQA Verified | — | 34.9% |
| Vectara Hallucination Rate | — | 10.8% |
Multimodal Grok 4.20 Multi-Agent leads
Grok 4.20 Multi-Agent: 39.8 (#44), Kimi K2.6: 31.8 (#95)
| Benchmark | Grok 4.20 Multi-Agent | Kimi K2.6 |
|---|---|---|
| LMArena Vision | 1259 | 1282 |
| Blueprint-Bench 2 | — | 3.9% |
| Furniture Assembly | — | 21.7% |
| LMArena Document | — | 1451 |
Multilingual Too close to call
Grok 4.20 Multi-Agent: 54.6 (#40), Kimi K2.6: 55.1 (#35)
| Benchmark | Grok 4.20 Multi-Agent | Kimi K2.6 |
|---|---|---|
| LMArena Non-English | 1440 | 1446 |
| LMArena Chinese | 1475 | 1521 |
| LMArena French | 1467 | 1471 |
| LMArena German | 1456 | 1449 |
| LMArena Japanese | 1405 | 1442 |
| LMArena Korean | 1416 | 1428 |
| LMArena Russian | 1457 | 1446 |
| LMArena Spanish | 1447 | 1463 |
Instruction Following Kimi K2.6 leads
Grok 4.20 Multi-Agent: 75.0 (#80), Kimi K2.6: 76.5 (#40)
| Benchmark | Grok 4.20 Multi-Agent | Kimi K2.6 |
|---|---|---|
| LMArena Instruction Following | 1420 | 1451 |
Long Context Kimi K2.6 leads
Grok 4.20 Multi-Agent: 43.7 (#84), Kimi K2.6: 44.9 (#50)
| Benchmark | Grok 4.20 Multi-Agent | Kimi K2.6 |
|---|---|---|
| LMArena Longer Query | 1430 | 1468 |
Writing & Preference Kimi K2.6 leads
Grok 4.20 Multi-Agent: 64.1 (#57), Kimi K2.6: 68.7 (#25)
| Benchmark | Grok 4.20 Multi-Agent | Kimi K2.6 |
|---|---|---|
| LMArena Text | 1450 | 1455 |
| LMArena Creative Writing | 1435 | 1434 |
| LMArena Multi-Turn | 1451 | 1453 |
| EQ-Bench Creative Writing | — | 1725 |
| EQ-Bench 4 | — | 1202 |
Frequently asked questions
Is Grok 4.20 Multi-Agent better than Kimi K2.6?
Kimi K2.6 is the stronger model overall, scoring 47.5 to 45.5 on the Noometry Index.
Which is cheaper, Grok 4.20 Multi-Agent or Kimi K2.6?
Grok 4.20 Multi-Agent is cheaper. It lists at $1.25 per million input tokens and $2.50 per million output tokens; Kimi K2.6 lists at $0.95 and $4.
Is Grok 4.20 Multi-Agent or Kimi K2.6 better for coding?
Kimi K2.6 scores higher on coding benchmarks: 50.5 versus 42.4 in the Noometry coding category.
Which has the bigger context window?
Grok 4.20 Multi-Agent does, with 1M tokens against 262K.
How many benchmarks do Grok 4.20 Multi-Agent and Kimi K2.6 share?
19 benchmarks have published results for both models. Grok 4.20 Multi-Agent has 20 scored results on Noometry and Kimi K2.6 has 51.