SShogo Bench

GPT-5.6 Sol vs Kimi K3

GPT-5.6 Sol leads Kimi K3 by 6.9 points on the Shogo Agent Index, but their ranges overlap, so treat it as close. GPT-5.6 Sol wins 18 and Kimi K3 wins 5 of 23 shared benchmarks. Kimi K3 is cheaper for agent workloads.

Add more models to this comparison →

Pillars

PillarGPT-5.6 SolKimi K3
Agentic coding7163
Long-horizon autonomy7265
Knowledge-work agents4642
Computer use5343
Security & context learning5853

Shared benchmarks

BenchmarkGPT-5.6 SolKimi K3
APEX-Agentsindex51.4%47.8%
DeepSWEindex72.7%68.5%
Vending-Bench 2index$9,619$5,165
FrontierSWEindex32.2%25.9%
PostTrainBenchindex31.0%31.5%
ARC-AGI97.5%94.5%
ARC-AGI-292.5%60.4%
Chess Puzzles68.0%37.0%
CritPt32.3%23.4%
DTBench96.0%91.2%
FrontierCode47.5%44.2%
FrontierMath Tier 482.9%39.0%
FrontierMath Tiers 1-389.1%72.2%
GPQA Diamond93.5%93.1%
LMCA59.2%52.7%
MirrorCode20.0%54.4%
OTIS Mock AIME100.0%97.2%
ProofBench83.0%87.0%
SciCode57.8%59.5%
SimpleBench71.7%60.7%
SimpleQA Verified69.7%50.6%
WebDev Arena16171674
WeirdML v315.0%7.5%

Specs and price

GPT-5.6 SolKimi K3
Agent Index61.8 (#14)54.9 (#26)
Agent $/M$1.85$1.39
Input / output $/M$4.00 / $20.00$3.00 / $15.00
Context1.1M1M
ReleasedAug 7, 2026Jul 16, 2026
WeightsClosedOpen
API hosts529

Don’t pick one. Route.

Shogo can run GPT-5.6 Sol 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