Model comparison
Gemma 4 31B vs Kimi K2.6
Head-to-head evidence from 26 shared benchmark results across 6 categories. Overall scores shown here use BenchLM's provisional ranking lane.
Verified leaderboard positions: Gemma 4 31B unranked; Kimi K2.6 #13
Evidence parity. Gemma 4 31B and Kimi K2.6 share 26 comparable benchmark results. 3 of 8 categories are comparable. 4 results are unique to Gemma 4 31B; 34 to Kimi K2.6.
Updated July 13, 2026- Shared results
- 26
- Gemma 4 31B only
- 4
- Kimi K2.6 only
- 34
- Comparable categories
- 3 / 8
Pick Kimi K2.6 if you want the stronger benchmark profile. Gemma 4 31B only becomes the better choice if knowledge is the priority or you want the cheaper token bill.
Confidence note. This is a partial-evidence comparison with 26 shared benchmark results across 6 evidence categories; 3 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.
Why this result
Kimi K2.6 is clearly ahead on the provisional aggregate, 74 to 61. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.
Kimi K2.6's sharpest advantage is in coding, where it averages 72.6 against 41.6. The single biggest benchmark swing on the page is HLE, 26.5% to 34.7%. Gemma 4 31B does hit back in knowledge, so the answer changes if that is the part of the workload you care about most.
Kimi K2.6 is also the more expensive model on tokens at $0.95 input / $4.00 output per 1M tokens, versus $0.00 input / $0.00 output per 1M tokens for Gemma 4 31B. That is roughly Infinityx on output cost alone.
Category breakdown
Exact category averages are shown below. Not measured means BenchLM does not have enough sourced public coverage for that model and category.
| Category | Gemma 4 31B | Δ | Kimi K2.6 |
|---|---|---|---|
| Coding | Gemma 4 31B41.6 | Margin→ 31.0 | Kimi K2.672.6 |
| Knowledge | Gemma 4 31B53.3 | Margin← 11.1 | Kimi K2.642.2 |
| Multimodal | Gemma 4 31B76.9 | Margin→ 2.9 | Kimi K2.679.8 |
| Agentic | Gemma 4 31BNot measured | MarginNo overlap | Kimi K2.673.5 |
| Math | Gemma 4 31BNot measured | MarginNo overlap | Kimi K2.667.1 |
Decisive benchmark drivers
The largest measured benchmark gaps in this matchup, with exact reported values.
More
- Source ↗
HLE
KnowledgeA 26.5%B 34.7%Winner: Kimi K2.6Δ 8.2HLE: Gemma 4 31B scored 26.5%; Kimi K2.6 scored 34.7%. Kimi K2.6 wins this benchmark. - Source ↗
GPQA
KnowledgeA 84.3%B 90.5%Winner: Kimi K2.6Δ 6.2GPQA: Gemma 4 31B scored 84.3%; Kimi K2.6 scored 90.5%. Kimi K2.6 wins this benchmark. - Source ↗
MMMU-Pro
MultimodalA 76.9%B 79.4%Winner: Kimi K2.6Δ 2.5MMMU-Pro: Gemma 4 31B scored 76.9%; Kimi K2.6 scored 79.4%. Kimi K2.6 wins this benchmark.
Operational comparison
Runtime and commercial metrics are compared only when both models have a complete sourced value.
| Metric | Gemma 4 31B | Kimi K2.6 | Comparison |
|---|---|---|---|
| Input / output priceUSD per 1M tokens | Gemma 4 31B$0 input / $0 output | Kimi K2.6$0.95 input / $4 output | Gemma 4 31B has the lower combined listed price. |
| Generation speedtokens per second | Gemma 4 31BNot available | Kimi K2.6Not available | A complete speed comparison is not available. |
| First-answer latencyseconds to first token | Gemma 4 31BNot available | Kimi K2.6Not available | A complete latency comparison is not available. |
| Context windowmaximum listed tokens | Gemma 4 31B256K | Kimi K2.6256K | Listed context windows are equal. |
Benchmark Deep Dive
Agentic22 benchmarks
| Benchmark | Gemma 4 31B | Kimi K2.6 | Result |
|---|---|---|---|
| AA Agentic IndexSource | 14.4% | 30.3% | Kimi K2.6 leads |
| Tau2-TelecomSource | 59.9% | 95.9% | Kimi K2.6 leads |
| GDPval-AASource | 15.0% | 34.5% | Kimi K2.6 leads |
| GDPval-AASource | 801 | 1190 | Kimi K2.6 leads |
| Gert LabsSource | 35.26% | 56.82% | Kimi K2.6 leads |
| AA EnterpriseOps-GymSource | 28.0% | 38.5% | Kimi K2.6 leads |
| AA Harvey LABSource | 0.0% | 0.0% | Tie |
| AA ITBenchSource | 37.3% | 31.2% | Gemma 4 31B leads |
| AA Tau3 BankingSource | 15.1% | 20.6% | Kimi K2.6 leads |
| Terminal-Bench 2.0Source | — | 66.7% | Not comparable |
| BrowseCompSource | — | 83.2% | Not comparable |
| OSWorld-VerifiedSource | — | 73.1% | Not comparable |
| ToolathlonSource | — | 50% | Not comparable |
| MCP AtlasSource | — | 55.9% | Not comparable |
| Claw-EvalSource | — | 62.3% | Not comparable |
| DeepSearchQASource | — | 92.5% | Not comparable |
| WideResearchSource | — | 80.8% | Not comparable |
| APEX-Agents-AASource | — | 28.5% | Not comparable |
| ResearchClawBenchSource | — | 18.0% | Not comparable |
| OSWorld 2.0Source | — | 4.6% | Not comparable |
| AA BriefcaseSource | — | 809 | Not comparable |
| AA AutomationBenchSource | — | 19.6% | Not comparable |
CodingKimi K2.6 wins15 benchmarks
| Benchmark | Gemma 4 31B | Kimi K2.6 | Result |
|---|---|---|---|
| SWE-RebenchSource | 41.6% | — | Not comparable |
| React Native EvalsSource | 75.2% | — | Not comparable |
| AA Coding IndexSource | 43.4% | 61.8% | Kimi K2.6 leads |
| Terminal-Bench HardSource | 36.4% | 43.9% | Kimi K2.6 leads |
| AA-SciCodeSource | 43.4% | 53.5% | Kimi K2.6 leads |
| SWE-bench VerifiedSource | — | 80.2% | Not comparable |
| LiveCodeBenchSource | — | 89.6% | Not comparable |
| LiveCodeBench v6Source | — | 89.6% | Not comparable |
| SWE-bench ProSource | — | 58.6% | Not comparable |
| SWE MultilingualSource | — | 76.7% | Not comparable |
| SciCodeSource | — | 52.2% | Not comparable |
| Terminal-Bench 2.0Source | — | 66.7% | Not comparable |
| Vibe Code BenchSource | — | 37.89% | Not comparable |
| cursorBench31Source | — | 47.6% | Not comparable |
| AA Terminal-Bench 2.1Source | — | 65.9% | Not comparable |
Reasoning2 benchmarks
KnowledgeGemma 4 31B wins12 benchmarks
| Benchmark | Gemma 4 31B | Kimi K2.6 | Result |
|---|---|---|---|
| GPQASource | 84.3% | 90.5% | Kimi K2.6 leads |
| MMLU-ProSource | 85.2% | — | Not comparable |
| HLESource | 26.5% | 34.7% | Kimi K2.6 leads |
| HLE w/o toolsSource | 19.5% | — | Not comparable |
| Artificial Analysis Intelligence IndexSource | 29.4% | 44.2% | Kimi K2.6 leads |
| AA-GPQA DiamondSource | 85.7% | 91.1% | Kimi K2.6 leads |
| AA-HLESource | 22.7% | 35.9% | Kimi K2.6 leads |
| AA-Omniscience IndexSource | -45.4% | 6.4% | Kimi K2.6 leads |
| AA-Omniscience AccuracySource | 19.9% | 32.8% | Kimi K2.6 leads |
| AA-Omniscience Hallucination RateSource | 81.6% | 39.3% | Kimi K2.6 leads |
| AA Openness IndexSource | 38.9% | 33.3% | Gemma 4 31B leads |
| GPQA-DSource | — | 90.5% | Not comparable |
Math5 benchmarks
MultimodalKimi K2.6 wins7 benchmarks
| Benchmark | Gemma 4 31B | Kimi K2.6 | Result |
|---|---|---|---|
| MMMU-ProSource | 76.9% | 79.4% | Kimi K2.6 leads |
| AA-MMMU-ProSource | 73.4% | 79.4% | Kimi K2.6 leads |
| MMMU-Pro w/ PythonSource | — | 80.1% | Not comparable |
| CharXivSource | — | 80.4% | Not comparable |
| MathVisionSource | — | 87.4% | Not comparable |
| V*Source | — | 96.9% | Not comparable |
| Design Arena WebsiteSource | — | 1318 | Not comparable |
Inst. Following1 benchmarks
| Benchmark | Gemma 4 31B | Kimi K2.6 | Result |
|---|---|---|---|
| AA-IFBenchSource | 75.6% | 76.0% | Kimi K2.6 leads |
Frequently Asked Questions (4)
Which is better, Gemma 4 31B or Kimi K2.6?
Kimi K2.6 is ahead on BenchLM's provisional leaderboard, 74 to 61. The biggest single separator in this matchup is HLE, where the scores are 26.5% and 34.7%.
Which is better for knowledge tasks, Gemma 4 31B or Kimi K2.6?
Gemma 4 31B has the edge for knowledge tasks in this comparison, averaging 53.3 versus 42.2. Inside this category, AA-Omniscience Index is the benchmark that creates the most daylight between them.
Which is better for coding, Gemma 4 31B or Kimi K2.6?
Kimi K2.6 has the edge for coding in this comparison, averaging 72.6 versus 41.6. Inside this category, AA Coding Index is the benchmark that creates the most daylight between them.
Which is better for multimodal and grounded tasks, Gemma 4 31B or Kimi K2.6?
Kimi K2.6 has the edge for multimodal and grounded tasks in this comparison, averaging 79.8 versus 76.9. Inside this category, AA-MMMU-Pro is the benchmark that creates the most daylight between them.
Self-host vs API cost
Estimates at 50,000 req/day · 1000 tokens/req average.
Related Comparisons
Explore More
The AI models change fast. We track them for you.
A weekly brief for engineers and researchers covering new models, ranking shifts, and pricing changes.
Free. No spam. Unsubscribe anytime.