DeepSeek V4.1 Flash vs Kimi K3
Kimi K3 leads DeepSeek V4.1 Flash by 3.7 points on the Shogo Agent Index, but their ranges overlap, so treat it as close. DeepSeek V4.1 Flash wins 5 and Kimi K3 wins 13 of 19 shared benchmarks. DeepSeek V4.1 Flash is cheaper for agent workloads.
Add more models to this comparison →Pillars
| Pillar | DeepSeek V4.1 Flash | Kimi K3 |
|---|---|---|
| Agentic coding | 57 | 63 |
| Long-horizon autonomy | 62 | 65 |
| Knowledge-work agents | 39 | 42 |
| Computer use | 40 | 43 |
| Security & context learning | 51 | 53 |
Shared benchmarks
| Benchmark | DeepSeek V4.1 Flash | Kimi K3 |
|---|---|---|
| APEX-Agentsindex | 39.5% | 47.8% |
| ARC-AGI | 94.5% | 94.5% |
| ARC-AGI-2 | 72.9% | 60.4% |
| Chess Puzzles | 48.0% | 37.0% |
| CritPt | 14.3% | 23.4% |
| DTBench | 89.9% | 91.2% |
| FrontierCode | 43.9% | 44.2% |
| FrontierMath Tier 4 | 26.8% | 39.0% |
| FrontierMath Tiers 1-3 | 67.4% | 72.2% |
| GPQA Diamond | 89.8% | 93.1% |
| LMCA | 47.0% | 52.7% |
| MirrorCode | 64.4% | 54.4% |
| OTIS Mock AIME | 98.3% | 97.2% |
| ProofBench | 54.0% | 87.0% |
| SciCode | 51.9% | 59.5% |
| SimpleBench | 66.7% | 60.7% |
| SimpleQA Verified | 42.9% | 50.6% |
| WebDev Arena | 1614 | 1674 |
| WeirdML v3 | 5.9% | 7.5% |
Specs and price
| DeepSeek V4.1 Flash | Kimi K3 | |
|---|---|---|
| Agent Index | 51.2 (#35) | 54.9 (#26) |
| Agent $/M | $0.12 | $1.39 |
| Input / output $/M | $0.30 / $1.20 | $3.00 / $15.00 |
| Context | 1M | 1M |
| Released | Sep 10, 2026 | Jul 16, 2026 |
| Weights | Open | Open |
| API hosts | 39 | 29 |
Don’t pick one. Route.
Shogo can run DeepSeek V4.1 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