Skip to content
gentic.news — AI News Intelligence Platform
Screen-level OS controlOpen Source#4 of 15 in category

Kimi K2.6

Moonshot AI · Launched Apr 2026

Moonshot AI's 1T-param MoE (32B active) built for long-horizon agentic coding (up to 13h continuous) with agent swarm scaling to 300 sub-agents. Leads SWE-Bench Pro at 58.6%, ~1/4 the cost of Claude Opus 4.6.

Visit Kimi K2.6API: 0.60/2.75 per M tokens
5
Benchmarks scored
89.6
Peak score
0
Article mentions
Yes
Open source

Benchmark performance

OSWorld-Verified

Real desktop workflows across browser, files, office apps. 369 tasks (361 without Google Drive). Human expert baseline 72.4%. Current SOTA: Holo3-35B-A3B (H Company) at 80.4%.

73.1
Gap to SOTA: -7.3pp (held by Holo3-35B-A3B)Human: 72.4%Benchmark docs →

Other screen-level os control agents

The 15 agents in this category, ranked by peak benchmark.

AgentMakerLaunchPeakPricing
Claude Sonnet 4.6Anthropic2026-021470.0API: 3/15 per M tokens
Kimi K2.5OSSMoonshot AI2026-011410.0API pay-as-you-go
Claude Computer UseAnthropic2024-1092.1Claude API — input $5/M, output $25/M
Holo3-35B-A3BH Company2026-0480.4H Company enterprise
Claude Sonnet 4.5Anthropic2025-0962.9Legacy Anthropic API
Seed-1.8ByteDance Seed2025-1261.9Doubao ecosystem
EvoCUA-20260105Meituan LongCat2026-0156.7Research
GUI-Owl-1.5 32BOSSAlibaba Tongyi Lab2026-0355.4Free (OSS)
DeepMiner-Mano-72BMininglamp Technology2025-1053.9Research
UI-TARS-2ByteDance Seed2025-1053.1ByteDance ecosystem

Quick facts

Type
Screen-level OS control
Maker
Moonshot AI
Launch
2026-04-20
Open source
Yes
Pricing
API: 0.60/2.75 per M tokens
Benchmarks scored
5
Article mentions
0
Rank in category
#4 of 15
Kimi K2.6 — Computer Use Agent | gentic.news