Model comparison
Gemma 3 4B vs Kimi K2 (Jul 2025)
Kimi K2 (Jul 2025) is the stronger model overall, scoring 40.9 to 27.9 on the Noometry Index. Gemma 3 4B costs 20× less per token, which makes it the better buy when Kimi K2 (Jul 2025)'s lead doesn't matter for your workload.
Last verified . 17 shared benchmarks.
GoogleSummary
- They share 17 benchmarks with published results for both. Gemma 3 4B scores higher in 0 categories and Kimi K2 (Jul 2025) in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Kimi K2 (Jul 2025) leads 36.9 to 11.8.
- The biggest single-benchmark swing is Berkeley Function Calling Leaderboard: 19.6% for Gemma 3 4B and 59.1% for Kimi K2 (Jul 2025).
- Gemma 3 4B is cheaper at $0.04 / $0.08 per million input/output tokens, against $0.57 / $2.30 for Kimi K2 (Jul 2025).
- Kimi K2 (Jul 2025) accepts more context: 262K tokens versus 131K.
Side by side
| Gemma 3 4B | Kimi K2 (Jul 2025) | |
|---|---|---|
| Provider | Moonshot AI | |
| Noometry Index | 27.9 | 40.9 |
| Released | 2025-03-12 | 2025-07-12 |
| Weights | Open | Open |
| Context window | 131K | 262K |
| Max output | 4K | 262K |
| Input $ / M tokens | $0.04 | $0.57 |
| Output $ / M tokens | $0.08 | $2.30 |
| Results tracked | 22 | 42 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K2 (Jul 2025) leads
Gemma 3 4B: 35.3 (#214), Kimi K2 (Jul 2025): 43.1 (#83)
| Benchmark | Gemma 3 4B | Kimi K2 (Jul 2025) |
|---|---|---|
| LMArena Coding | 1230 | 1399 |
| SWE-bench Verified (bash only) | — | 63.4% |
| Aider Polyglot | — | 59.1% |
| GSO | — | 4.9% |
| WeirdML | — | 42.8% |
| ALE-Bench | — | 597.5 |
Agentic & Tool Use Kimi K2 (Jul 2025) leads
Gemma 3 4B: 20.7 (#134), Kimi K2 (Jul 2025): 32.2 (#59)
| Benchmark | Gemma 3 4B | Kimi K2 (Jul 2025) |
|---|---|---|
| Berkeley Function Calling Leaderboard | 19.6% | 59.1% |
| Terminal-Bench | — | 35.7% |
| METR Time Horizons | — | 59.2% |
Reasoning Kimi K2 (Jul 2025) leads
Gemma 3 4B: 12.7 (#330), Kimi K2 (Jul 2025): 22.7 (#173)
| Benchmark | Gemma 3 4B | Kimi K2 (Jul 2025) |
|---|---|---|
| Kagi LLM Benchmark | 25.2% | 64.4% |
| LMArena Hard Prompts | 1253 | 1384 |
| Epoch Capabilities Index | 116.02 | 146.02 |
| SimpleBench | — | 26.3% |
| Chess Puzzles | 0% | — |
| DTBench | 50.9% | — |
| LMCA | 2.8% | — |
| ForecastBench | — | 60.2 |
Math Kimi K2 (Jul 2025) leads
Gemma 3 4B: 15.8 (#286), Kimi K2 (Jul 2025): 40.1 (#81)
| Benchmark | Gemma 3 4B | Kimi K2 (Jul 2025) |
|---|---|---|
| LMArena Math | 1239 | 1396 |
| OTIS Mock AIME 2024-2025 | 7.5% | — |
| Omni-MATH | — | 65.4% |
| FrontierMath (Feb 2025 set) | — | 21.4% |
| FrontierMath Tier 4 (v1) | — | 0% |
Knowledge Kimi K2 (Jul 2025) leads
Gemma 3 4B: 11.8 (#294), Kimi K2 (Jul 2025): 36.9 (#153)
| Benchmark | Gemma 3 4B | Kimi K2 (Jul 2025) |
|---|---|---|
| Vectara Hallucination Rate | 6.4% | 17.9% |
| LMArena Expert | 1223 | 1363 |
| GPQA Diamond | 23.2% | — |
| MMLU-Pro | — | 81.9% |
| Confabulations | — | 20.4% |
| GPQA (HELM) | — | 65.3% |
Multilingual Kimi K2 (Jul 2025) leads
Gemma 3 4B: 42.7 (#189), Kimi K2 (Jul 2025): 49.8 (#125)
| Benchmark | Gemma 3 4B | Kimi K2 (Jul 2025) |
|---|---|---|
| LMArena Non-English | 1273 | 1372 |
| LMArena German | 1281 | 1386 |
| LMArena Russian | 1294 | 1384 |
| LMArena Chinese | — | 1416 |
| LMArena French | — | 1379 |
| LMArena Japanese | — | 1349 |
| LMArena Korean | — | 1324 |
| LMArena Spanish | — | 1383 |
Instruction Following Kimi K2 (Jul 2025) leads
Gemma 3 4B: 65.4 (#219), Kimi K2 (Jul 2025): 71.3 (#151)
| Benchmark | Gemma 3 4B | Kimi K2 (Jul 2025) |
|---|---|---|
| LMArena Instruction Following | 1239 | 1347 |
| IFEval | — | 85% |
Long Context Kimi K2 (Jul 2025) leads
Gemma 3 4B: 38.6 (#189), Kimi K2 (Jul 2025): 41.2 (#140)
| Benchmark | Gemma 3 4B | Kimi K2 (Jul 2025) |
|---|---|---|
| LMArena Longer Query | 1274 | 1353 |
| Fiction.LiveBench | — | 66.7% |
| CL-bench | — | 17.6% |
Writing & Preference Kimi K2 (Jul 2025) leads
Gemma 3 4B: 42.2 (#234), Kimi K2 (Jul 2025): 62.5 (#76)
| Benchmark | Gemma 3 4B | Kimi K2 (Jul 2025) |
|---|---|---|
| LMArena Text | 1291 | 1380 |
| LMArena Creative Writing | 1270 | 1349 |
| EQ-Bench Creative Writing | 1068 | 1666 |
| LMArena Multi-Turn | 1255 | 1371 |
| Short-Story Creative Writing | — | 85.6% |
| WildBench | — | 86.2% |
Frequently asked questions
Is Gemma 3 4B better than Kimi K2 (Jul 2025)?
Kimi K2 (Jul 2025) is the stronger model overall, scoring 40.9 to 27.9 on the Noometry Index. Gemma 3 4B costs 20× less per token, which makes it the better buy when Kimi K2 (Jul 2025)'s lead doesn't matter for your workload.
Which is cheaper, Gemma 3 4B or Kimi K2 (Jul 2025)?
Gemma 3 4B is cheaper. It lists at $0.04 per million input tokens and $0.08 per million output tokens; Kimi K2 (Jul 2025) lists at $0.57 and $2.30.
Is Gemma 3 4B or Kimi K2 (Jul 2025) better for coding?
Kimi K2 (Jul 2025) scores higher on coding benchmarks: 43.1 versus 35.3 in the Noometry coding category.
Which has the bigger context window?
Kimi K2 (Jul 2025) does, with 262K tokens against 131K.
How many benchmarks do Gemma 3 4B and Kimi K2 (Jul 2025) share?
17 benchmarks have published results for both models. Gemma 3 4B has 22 scored results on Noometry and Kimi K2 (Jul 2025) has 42.