Gemini 3.8 Flash vs Kimi K3
Gemini 3.8 Flash leads Kimi K3 by 1.3 points on the Shogo Agent Index, but their ranges overlap, so treat it as close. Gemini 3.8 Flash wins 11 and Kimi K3 wins 11 of 22 shared benchmarks. Gemini 3.8 Flash is cheaper for agent workloads.
Add more models to this comparison →Pillars
| Pillar | Gemini 3.8 Flash | Kimi K3 |
|---|---|---|
| Agentic coding | 63 | 63 |
| Long-horizon autonomy | 64 | 65 |
| Knowledge-work agents | 47 | 42 |
| Computer use | 44 | 43 |
| Security & context learning | 53 | 53 |
Shared benchmarks
| Benchmark | Gemini 3.8 Flash | Kimi K3 |
|---|---|---|
| APEX-Agentsindex | 64.3% | 47.8% |
| DeepSWEindex | 73.8% | 68.5% |
| Vending-Bench 2index | $5,094 | $5,165 |
| FrontierSWEindex | 19.6% | 25.9% |
| ARC-AGI | 98.5% | 94.5% |
| ARC-AGI-2 | 89.2% | 60.4% |
| Chess Puzzles | 58.0% | 37.0% |
| CritPt | 18.3% | 23.4% |
| DTBench | 95.7% | 91.2% |
| FrontierCode | 41.2% | 44.2% |
| FrontierMath Tier 4 | 22.0% | 39.0% |
| FrontierMath Tiers 1-3 | 68.4% | 72.2% |
| GPQA Diamond | 95.4% | 93.1% |
| LMCA | 52.9% | 52.7% |
| MirrorCode | 38.9% | 54.4% |
| OTIS Mock AIME | 98.9% | 97.2% |
| ProofBench | 48.0% | 87.0% |
| SciCode | 56.6% | 59.5% |
| SimpleBench | 82.4% | 60.7% |
| SimpleQA Verified | 69.7% | 50.6% |
| WebDev Arena | 1568 | 1674 |
| WeirdML v3 | 7.4% | 7.5% |
Specs and price
| Gemini 3.8 Flash | Kimi K3 | |
|---|---|---|
| Agent Index | 56.2 (#21) | 54.9 (#26) |
| Agent $/M | $0.35 | $1.39 |
| Input / output $/M | $0.75 / $3.75 | $3.00 / $15.00 |
| Context | 1M | 1M |
| Released | Sep 2, 2026 | Jul 16, 2026 |
| Weights | Closed | Open |
| API hosts | 6 | 29 |
Don’t pick one. Route.
Shogo can run Gemini 3.8 Flash and Kimi K3 side by side and send each step of an agent to the one that’s cheaper for it.
Try both in Shogo