Model comparison
Gemma 3 4B vs Kimi K2.6
Kimi K2.6 is the stronger model overall, scoring 47.7 to 28.1 on the Noometry Index. Gemma 3 4B costs 34× less per token, which makes it the better buy when Kimi K2.6's lead doesn't matter for your workload.
Last verified . 20 shared benchmarks.
GoogleSummary
- They share 20 benchmarks with published results for both. Gemma 3 4B scores higher in 0 categories and Kimi K2.6 in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Kimi K2.6 leads 54.0 to 11.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 7.5% for Gemma 3 4B and 96.1% for Kimi K2.6.
- Gemma 3 4B is cheaper at $0.04 / $0.08 per million input/output tokens, against $0.95 / $4 for Kimi K2.6.
- Kimi K2.6 accepts more context: 262K tokens versus 131K.
Side by side
| Gemma 3 4B | Kimi K2.6 | |
|---|---|---|
| Provider | Moonshot AI | |
| Noometry Index | 28.1 | 47.7 |
| Released | 2025-03-12 | 2026-04-20 |
| Weights | Open | Open |
| Context window | 131K | 262K |
| Max output | 4K | 262K |
| Input $ / M tokens | $0.04 | $0.95 |
| Output $ / M tokens | $0.08 | $4 |
| Results tracked | 22 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K2.6 leads
Gemma 3 4B: 35.9 (#215), Kimi K2.6: 50.7 (#43)
| Benchmark | Gemma 3 4B | Kimi K2.6 |
|---|---|---|
| LMArena Coding | 1230 | 1488 |
| SWE-bench Verified | — | 76.7% |
| LMArena WebDev | — | 1509 |
| SciCode | — | 53.5% |
| WeirdML | — | 55.9% |
| ALE-Bench | — | 1,093 |
Agentic & Tool Use Kimi K2.6 leads
Gemma 3 4B: 20.9 (#142), Kimi K2.6: 21.9 (#137)
| Benchmark | Gemma 3 4B | Kimi K2.6 |
|---|---|---|
| Berkeley Function Calling Leaderboard | 19.6% | — |
| OSWorld 2.0 | — | 4.6% |
| ExploitBench | — | 18.4% |
| GBAEval | — | 0.9% |
| GDP.pdf | — | 12% |
| Vending-Bench 2 | — | 6,205 |
Reasoning Kimi K2.6 leads
Gemma 3 4B: 13.2 (#335), Kimi K2.6: 40.5 (#55)
| Benchmark | Gemma 3 4B | Kimi K2.6 |
|---|---|---|
| Chess Puzzles | 0% | 26% |
| LMArena Hard Prompts | 1253 | 1470 |
| DTBench | 50.9% | 90.9% |
| LMCA | 2.8% | 37.3% |
| Epoch Capabilities Index | 116.02 | 151.05 |
| Kagi LLM Benchmark | 25.2% | — |
| NYT Connections (extended) | — | 87.2% |
| CritPt | — | 8% |
| EBR-Bench | — | 2.4% |
| Mystery Game Puzzles | — | 18% |
Math Kimi K2.6 leads
Gemma 3 4B: 16.8 (#292), Kimi K2.6: 57.0 (#41)
| Benchmark | Gemma 3 4B | Kimi K2.6 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 7.5% | 96.1% |
| LMArena Math | 1239 | 1475 |
| FrontierMath (Tiers 1-3) | — | 57.2% |
| FrontierMath Tier 4 | — | 25.6% |
| MathArena Final-Answer Competitions | — | 72.9% |
| ProofBench | — | 16% |
| FrontierMath (Feb 2025 set) | — | 39% |
| FrontierMath Tier 4 (v1) | — | 14.6% |
Knowledge Kimi K2.6 leads
Gemma 3 4B: 11.8 (#299), Kimi K2.6: 54.0 (#54)
| Benchmark | Gemma 3 4B | Kimi K2.6 |
|---|---|---|
| GPQA Diamond | 23.2% | 90.8% |
| Vectara Hallucination Rate | 6.4% | 10.8% |
| LMArena Expert | 1223 | 1491 |
| SimpleQA Verified | — | 34.9% |
Multimodal Not comparable
Gemma 3 4B: —, Kimi K2.6: 31.6 (#103)
| Benchmark | Gemma 3 4B | Kimi K2.6 |
|---|---|---|
| LMArena Vision | — | 1283 |
| Blueprint-Bench 2 | — | 3.9% |
| Furniture Assembly | — | 21.7% |
| LMArena Document | — | 1451 |
Multilingual Kimi K2.6 leads
Gemma 3 4B: 42.5 (#194), Kimi K2.6: 54.9 (#37)
| Benchmark | Gemma 3 4B | Kimi K2.6 |
|---|---|---|
| LMArena Non-English | 1273 | 1446 |
| LMArena German | 1281 | 1450 |
| LMArena Russian | 1294 | 1446 |
| LMArena Chinese | — | 1521 |
| LMArena French | — | 1471 |
| LMArena Japanese | — | 1443 |
| LMArena Korean | — | 1427 |
| LMArena Spanish | — | 1464 |
Instruction Following Kimi K2.6 leads
Gemma 3 4B: 65.2 (#225), Kimi K2.6: 76.3 (#43)
| Benchmark | Gemma 3 4B | Kimi K2.6 |
|---|---|---|
| LMArena Instruction Following | 1239 | 1451 |
Long Context Kimi K2.6 leads
Gemma 3 4B: 38.7 (#194), Kimi K2.6: 44.9 (#52)
| Benchmark | Gemma 3 4B | Kimi K2.6 |
|---|---|---|
| LMArena Longer Query | 1273 | 1468 |
Writing & Preference Kimi K2.6 leads
Gemma 3 4B: 42.0 (#239), Kimi K2.6: 68.5 (#26)
| Benchmark | Gemma 3 4B | Kimi K2.6 |
|---|---|---|
| LMArena Text | 1291 | 1455 |
| LMArena Creative Writing | 1271 | 1434 |
| EQ-Bench Creative Writing | 1068 | 1725 |
| LMArena Multi-Turn | 1255 | 1453 |
| EQ-Bench 4 | — | 1202 |
Frequently asked questions
Is Gemma 3 4B better than Kimi K2.6?
Kimi K2.6 is the stronger model overall, scoring 47.7 to 28.1 on the Noometry Index. Gemma 3 4B costs 34× less per token, which makes it the better buy when Kimi K2.6's lead doesn't matter for your workload.
Which is cheaper, Gemma 3 4B or Kimi K2.6?
Gemma 3 4B is cheaper. It lists at $0.04 per million input tokens and $0.08 per million output tokens; Kimi K2.6 lists at $0.95 and $4.
Is Gemma 3 4B or Kimi K2.6 better for coding?
Kimi K2.6 scores higher on coding benchmarks: 50.7 versus 35.9 in the Noometry coding category.
Which has the bigger context window?
Kimi K2.6 does, with 262K tokens against 131K.
How many benchmarks do Gemma 3 4B and Kimi K2.6 share?
20 benchmarks have published results for both models. Gemma 3 4B has 22 scored results on Noometry and Kimi K2.6 has 51.