Model comparison
Gemma 4 31B IT vs Kimi K3
Kimi K3 is the stronger model overall, scoring 59.5 to 43.5 on the Noometry Index. Gemma 4 31B IT costs 39× less per token, which makes it the better buy when Kimi K3's lead doesn't matter for your workload.
Last verified . 30 shared benchmarks.
GoogleSummary
- They share 30 benchmarks with published results for both. Gemma 4 31B IT scores higher in 1 category and Kimi K3 in 8 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Kimi K3 leads 63.0 to 27.2.
- The biggest single-benchmark swing is Surface Evolver Bench: 30.6% for Gemma 4 31B IT and 95% for Kimi K3.
- Gemma 4 31B IT is cheaper at $0.09 / $0.34 per million input/output tokens, against $3 / $15 for Kimi K3.
- Kimi K3 accepts more context: 1.05M tokens versus 262K.
Side by side
| Gemma 4 31B IT | Kimi K3 | |
|---|---|---|
| Provider | Moonshot AI | |
| Noometry Index | 43.5 | 59.5 |
| Released | 2026-04-02 | 2026-07-16 |
| Weights | Open | Open |
| Context window | 262K | 1.05M |
| Max output | 33K | 1.05M |
| Input $ / M tokens | $0.09 | $3 |
| Output $ / M tokens | $0.34 | $15 |
| Results tracked | 35 | 53 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K3 leads
Gemma 4 31B IT: 42.3 (#108), Kimi K3: 61.0 (#10)
| Benchmark | Gemma 4 31B IT | Kimi K3 |
|---|---|---|
| LMArena WebDev | 1366 | 1654 |
| SciCode | 43.4% | 59.5% |
| WeirdML | 52.3% | 82.6% |
| LMArena Coding | 1459 | 1508 |
| ALE-Bench | 925.5 | 1,524 |
| DeepSWE | — | 68.5% |
| FrontierCode | — | 44.2% |
| FrontierSWE | — | 25.9% |
Agentic & Tool Use Not comparable
Gemma 4 31B IT: —, Kimi K3: 41.8 (#20)
| Benchmark | Gemma 4 31B IT | Kimi K3 |
|---|---|---|
| APEX-Agents | — | 50.6% |
| τ²-bench Banking | — | 37.1% |
| PostTrainBench | — | 32% |
| GBAEval | — | 48.3% |
| GDP.pdf | — | 19% |
| Vending-Bench 2 | — | 5,165 |
Reasoning Kimi K3 leads
Gemma 4 31B IT: 27.2 (#122), Kimi K3: 63.0 (#17)
| Benchmark | Gemma 4 31B IT | Kimi K3 |
|---|---|---|
| NYT Connections (extended) | 70.6% | 93.6% |
| CritPt | 1.4% | 23.4% |
| Chess Puzzles | 5% | 39% |
| LMArena Hard Prompts | 1448 | 1496 |
| DTBench | 82.7% | 91.2% |
| LMCA | 39.3% | 52.7% |
| Surface Evolver Bench | 30.6% | 95% |
| Epoch Capabilities Index | 142.74 | 157.45 |
| ARC-AGI-2 | — | 60.4% |
| SimpleBench | — | 60.7% |
| Kagi LLM Benchmark | 63.5% | — |
| ARC-AGI-1 | — | 94.5% |
| Thematic Generalization | 53% | — |
| Mystery Game Puzzles | — | 26% |
| ForecastBench | — | 61.1 |
Math Kimi K3 leads
Gemma 4 31B IT: 43.2 (#81), Kimi K3: 74.2 (#16)
| Benchmark | Gemma 4 31B IT | Kimi K3 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 73.3% | 97.2% |
| LMArena Math | 1465 | 1491 |
| FrontierMath (Tiers 1-3) | — | 72.2% |
| FrontierMath Tier 4 | — | 39% |
| MathArena Final-Answer Competitions | — | 87.8% |
| ProofBench | — | 87% |
Knowledge Kimi K3 leads
Gemma 4 31B IT: 37.9 (#151), Kimi K3: 63.2 (#21)
| Benchmark | Gemma 4 31B IT | Kimi K3 |
|---|---|---|
| GPQA Diamond | 75.8% | 93.1% |
| SimpleQA Verified | 10.4% | 50.6% |
| LMArena Expert | 1465 | 1521 |
| Vectara Hallucination Rate | 7.4% | — |
Multimodal Gemma 4 31B IT leads
Gemma 4 31B IT: 41.6 (#34), Kimi K3: 37.8 (#70)
| Benchmark | Gemma 4 31B IT | Kimi K3 |
|---|---|---|
| LMArena Vision | 1277 | — |
| Blueprint-Bench 2 | — | 29.5% |
| Furniture Assembly | — | 34.2% |
| LMArena Document | 1425 | — |
Multilingual Kimi K3 leads
Gemma 4 31B IT: 53.8 (#57), Kimi K3: 56.3 (#21)
| Benchmark | Gemma 4 31B IT | Kimi K3 |
|---|---|---|
| LMArena Non-English | 1431 | 1466 |
| LMArena Chinese | 1476 | 1529 |
| LMArena French | 1435 | 1491 |
| LMArena Russian | 1460 | 1482 |
| LMArena Spanish | 1444 | 1472 |
| LMArena German | — | 1488 |
| LMArena Japanese | — | 1487 |
| LMArena Korean | — | 1458 |
Instruction Following Kimi K3 leads
Gemma 4 31B IT: 75.5 (#61), Kimi K3: 77.7 (#14)
| Benchmark | Gemma 4 31B IT | Kimi K3 |
|---|---|---|
| LMArena Instruction Following | 1433 | 1483 |
Long Context Kimi K3 leads
Gemma 4 31B IT: 44.2 (#71), Kimi K3: 45.8 (#29)
| Benchmark | Gemma 4 31B IT | Kimi K3 |
|---|---|---|
| LMArena Longer Query | 1446 | 1494 |
Writing & Preference Kimi K3 leads
Gemma 4 31B IT: 60.5 (#96), Kimi K3: 76.6 (#4)
| Benchmark | Gemma 4 31B IT | Kimi K3 |
|---|---|---|
| LMArena Text | 1443 | 1476 |
| LMArena Creative Writing | 1415 | 1454 |
| EQ-Bench Creative Writing | 1368 | 2082 |
| EQ-Bench 4 | 1120 | 1339 |
| LMArena Multi-Turn | 1452 | 1488 |
Frequently asked questions
Is Gemma 4 31B IT better than Kimi K3?
Kimi K3 is the stronger model overall, scoring 59.5 to 43.5 on the Noometry Index. Gemma 4 31B IT costs 39× less per token, which makes it the better buy when Kimi K3's lead doesn't matter for your workload.
Which is cheaper, Gemma 4 31B IT or Kimi K3?
Gemma 4 31B IT is cheaper. It lists at $0.09 per million input tokens and $0.34 per million output tokens; Kimi K3 lists at $3 and $15.
Is Gemma 4 31B IT or Kimi K3 better for coding?
Kimi K3 scores higher on coding benchmarks: 61.0 versus 42.3 in the Noometry coding category.
Which has the bigger context window?
Kimi K3 does, with 1.05M tokens against 262K.
How many benchmarks do Gemma 4 31B IT and Kimi K3 share?
30 benchmarks have published results for both models. Gemma 4 31B IT has 35 scored results on Noometry and Kimi K3 has 53.