Gemini 4 Argon vs Kimi K3
Gemini 4 Argon leads Kimi K3 by 14.6 points on the Shogo Agent Index. Gemini 4 Argon wins 22 and Kimi K3 wins 0 of 22 shared benchmarks. Gemini 4 Argon is cheaper for agent workloads.
Add more models to this comparison →Pillars
| Pillar | Gemini 4 Argon | Kimi K3 |
|---|---|---|
| Agentic coding | 77 | 63 |
| Long-horizon autonomy | 78 | 65 |
| Knowledge-work agents | 59 | 42 |
| Computer use | 61 | 43 |
| Security & context learning | 61 | 53 |
Shared benchmarks
| Benchmark | Gemini 4 Argon | Kimi K3 |
|---|---|---|
| APEX-Agentsindex | 82.4% | 47.8% |
| Vending-Bench 2index | $13,718 | $5,165 |
| FrontierSWEindex | 55.0% | 25.9% |
| PostTrainBenchindex | 40.4% | 31.5% |
| ARC-AGI | 98.5% | 94.5% |
| ARC-AGI-2 | 91.3% | 60.4% |
| Chess Puzzles | 59.0% | 37.0% |
| CritPt | 27.1% | 23.4% |
| DTBench | 97.6% | 91.2% |
| FrontierCode | 47.5% | 44.2% |
| FrontierMath Tier 4 | 78.0% | 39.0% |
| FrontierMath Tiers 1-3 | 86.0% | 72.2% |
| GPQA Diamond | 96.4% | 93.1% |
| LMCA | 63.7% | 52.7% |
| MirrorCode | 58.9% | 54.4% |
| OTIS Mock AIME | 100.0% | 97.2% |
| ProofBench | 99.0% | 87.0% |
| SciCode | 61.8% | 59.5% |
| SimpleBench | 91.7% | 60.7% |
| SimpleQA Verified | 76.0% | 50.6% |
| WebDev Arena | 1680 | 1674 |
| WeirdML v3 | 20.3% | 7.5% |
Specs and price
| Gemini 4 Argon | Kimi K3 | |
|---|---|---|
| Agent Index | 69.5 (#6) | 54.9 (#26) |
| Agent $/M | $0.84 | $1.39 |
| Input / output $/M | $2.00 / $10.00 | $3.00 / $15.00 |
| Context | 1M | 1M |
| Released | Oct 13, 2026 | Jul 16, 2026 |
| Weights | Closed | Open |
| API hosts | 5 | 29 |
Don’t pick one. Route.
Shogo can run Gemini 4 Argon 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