MoonshotAI: Kimi K2.6 vs Google: Gemma 4 31B
Side-by-side comparison of specs, pricing, benchmark scores, and task rankings. Updated 2026-11-07.
| MoonshotAI: Kimi K2.6 | Google: Gemma 4 31B | |
|---|---|---|
| Vendor | moonshotai | |
| Quality Score | 100 | 100 |
| Benchmark Score | 76.7 | 53.7 |
| Input Price | $0.65/M | $0.20/M |
| Output Price | $3.41/M | $0.40/M |
| Context Window | 262,144 | 262,144 |
| Max Output | 235,929 | 16,384 |
| Tool Calling | ✓ | ✓ |
| Structured Output | ✓ | ✓ |
| Reasoning Mode | ✓ | ✓ |
| Vision | ✓ | ✓ |
| Audio | - | - |
| Benchmark Scores | ||
| ai_index | 73.0 | 48.4 |
| ai_index_agentic | 49.9 | 23.8 |
| ai_index_coding | 100.0 | 71.7 |
| eqbench | 63.5 | 57.5 |
The verdict
For a sample job of 1 million input tokens and 250,000 output tokens, Gemma 4 31B costs $0.30 and Kimi K2.6 costs $1.50, so Gemma 4 31B is about 5.0 times cheaper for the same work.
Both read up to 262,144 tokens at once, roughly 393 pages of text.
Only Gemma 4 31B offers video input.
On our blended benchmark score Kimi K2.6 leads 76.7 to 53.7.
Across the 19 tasks where they differ, Kimi K2.6 scores higher on 19 (SQL Generation, Code Review, Code Refactoring and others).
Pick Kimi K2.6 if you need benchmark performance or the stronger all-round task scores; pick Gemma 4 31B if you need lower cost or video input.
Costs use current list prices per million tokens; page counts assume about 0.75 words per token and 500 words per page.
Who wins by task?
| Task | MoonshotAI: Kimi K2.6 | Google: Gemma 4 31B |
|---|---|---|
| SQL Generation | 168 | 156 |
| Code Review | 164 | 152 |
| Code Refactoring | 160 | 151 |
| Bug Fixing | 174 | 160 |
| Unit Test Generation | 152 | 143 |
| Code Documentation | 140 | 137 |
| Regex Writing | 136 | 131 |
| CI/CD Pipelines | 142 | 135 |
| Frontend Component Design | 145 | 138 |
| Data Analysis | 166 | 152 |
| CSV / Spreadsheet Cleanup | 151 | 145 |
| ETL Scripting | 151 | 143 |
| JSON Extraction | 147 | 143 |
| Bulk Data Labeling | 134 | 133 |
| OCR / Document Parsing | 144 | 140 |
| Table Extraction from PDFs | 144 | 140 |
| Long-Document Summarization | 156 | 149 |
| Short-Form Summarization | 131 | 129 |
| Blog Post Writing | 139 | 133 |
Scores reflect capability match + benchmark data + pricing for each task. Methodology →