Comparing 2 models
Gemini 3.5 Flash vs Kimi K3
Benchmarks through
Verdict
Gemini 3.5 Flash places higher on 3 of the 3 tasks where both are ranked. Kimi K3 costs 7.8x less per input token. Only Kimi K3 has open weights for self-hosting.
- Choose Gemini 3.5 Flash for Translation (#7), Literary Translation (#7), Everyday Translation (#7).
- Choose Kimi K3 for you need to self-host.
Built from the published task rankings below; ranks come from each task's own weighted benchmarks, so a win on one task says nothing about another.
Measured head to head
Benchmarks both models were run on
| Evaluation | Gemini 3.5 Flash | Kimi K3 | Source |
|---|---|---|---|
| Design Arena Coding | 1256 Elo | 1352 Elo | source · 2026-11-15 |
| Design Arena Full-stack Agent | 1170 Elo | 1289 Elo | source · 2026-11-15 |
| Design Arena UI | 1265 Elo | 1342 Elo | source · 2026-11-15 |
| Design Arena Web-app Agent | 1177 Elo | 1273 Elo | source · 2026-11-15 |
| Design Arena Website | 1260 Elo | 1336 Elo | source · 2026-11-15 |
| FrontierMath Tiers 1-3 (Epoch AI) | 62.8% ± 2.9 | 72.2% ± 2.7 | source · 2026-06-10 |
| GPQA Diamond (Epoch AI) | 88.9% ± 2.2 | 93.1% ± 1.5 | source · 2026-08-06 |
| GPQA Diamond (Epoch AI) | 86.4% ± 2.4 | 93.1% ± 1.5 | source · 2026-07-15 |
| GPQA Diamond (Epoch AI) | 92.8% ± 1.6 | 93.1% ± 1.5 | source · 2026-05-22 |
| LiveBench Agentic Coding | 49.0% | 62.2% | source · 2026-10-18 |
| LiveBench Coding | 78.2% | 81.5% | source · 2026-10-18 |
| LiveBench Data Analysis | 64.9% | 78.7% | source · 2026-10-18 |
| LiveBench Instruction Following | 75.6% | 71.4% | source · 2026-10-18 |
| LiveBench Language | 84.6% | 85.5% | source · 2026-10-18 |
| LiveBench Math | 88.2% | 84.4% | source · 2026-10-18 |
| LiveBench Reasoning | 82% | 90.7% | source · 2026-10-18 |
| LMArena Creative Writing | 1469 Elo | 1453 Elo | source · 2026-11-11 |
| LMArena Instruction Following | 1467 Elo | 1484 Elo | source · 2026-11-11 |
| LMArena Long Query | 1484 Elo | 1490 Elo | source · 2026-11-11 |
| LMArena Math | 1504 Elo | 1496 Elo | source · 2026-11-11 |
| LMArena Text | 1478 Elo | 1487 Elo | source · 2026-11-11 |
| LMArena Text Factuality | 1478 Elo | 1471 Elo | source · 2026-11-11 |
| LMArena WebDev | 1500 Elo | 1658 Elo | source · 2026-11-11 |
| LMArena WebDev Image-to-Web | 1500 Elo | 1586 Elo | source · 2026-11-10 |
| OTIS Mock AIME (Epoch AI) | 88.9% ± 4.7 | 97.2% ± 1.1 | source · 2026-08-06 |
| OTIS Mock AIME (Epoch AI) | 80% ± 6 | 97.2% ± 1.1 | source · 2026-07-15 |
| OTIS Mock AIME (Epoch AI) | 95.6% ± 2.7 | 97.2% ± 1.1 | source · 2026-05-25 |
| SimpleQA Verified (Epoch AI) | 66.2% ± 1.5 | 50.6% ± 1.6 | source · 2026-08-27 |
Values are as published by each evaluator, in its own units; rows from different evaluators are not on a common scale.
task evidence
How they place when the question changes
Lower ranks are better. “Best” marks the strongest option in this shortlist for that task.