SShogo Bench

Grok 4.5 vs Kimi K3

Kimi K3 leads Grok 4.5 by 5.5 points on the Shogo Agent Index, but their ranges overlap, so treat it as close. Grok 4.5 wins 5 and Kimi K3 wins 16 of 21 shared benchmarks. Grok 4.5 is cheaper for agent workloads.

Add more models to this comparison →

Pillars

PillarGrok 4.5Kimi K3
Agentic coding5463
Long-horizon autonomy5965
Knowledge-work agents4242
Computer use3643
Security & context learning4853

Shared benchmarks

BenchmarkGrok 4.5Kimi K3
APEX-Agentsindex56.2%47.8%
DeepSWEindex53.8%68.5%
Vending-Bench 2index$3,887$5,165
PostTrainBenchindex22.6%31.5%
ARC-AGI87.2%94.5%
ARC-AGI-252.6%60.4%
Chess Puzzles35.0%37.0%
CritPt15.4%23.4%
DTBench96.5%91.2%
FrontierCode42.4%44.2%
FrontierMath Tier 424.4%39.0%
FrontierMath Tiers 1-357.2%72.2%
GPQA Diamond93.4%93.1%
LMCA45.2%52.7%
MirrorCode32.2%54.4%
OTIS Mock AIME97.8%97.2%
ProofBench31.0%87.0%
SciCode55.0%59.5%
SimpleBench70.0%60.7%
SimpleQA Verified48.3%50.6%
WebDev Arena15551674

Specs and price

Grok 4.5Kimi K3
Agent Index49.4 (#36)54.9 (#26)
Agent $/M$0.81$1.39
Input / output $/M$2.00 / $6.00$3.00 / $15.00
Context500K1M
ReleasedJul 8, 2026Jul 16, 2026
WeightsClosedOpen
API hosts429

Don’t pick one. Route.

Shogo can run Grok 4.5 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