Cognition floods Devin with GPT-6 models and cuts task costs 61%, gives away 50 Max plans

EricBuess · x · 2026-09-23

Cognition announced a wave of new models in Devin to celebrate fresh launches: GPT-6 Astra/Sol/Luna, Claude Opus 5.5 and Fable 5.1, Gemini 3.8 Flash, Grok 4.7, Kimi K3, DeepSeek V4.1 Flash, GLM-5.3 Flash, plus Inkling and SWE-2 (free until October 15), with cloud agents now on Linux, macOS, and Windows via a new Fusion Frontier harness.

Official benchmarks show steep cost cuts: on FrontierCode 1.1, GPT-6 Sol matches GPT-5.6 Sol's score at 61% lower cost per task, while GPT-6 Luna beats GPT-5.6 Luna at roughly a quarter of the cost. At under $0.10 per task, it is the cheapest model on the leaderboard.

Cognition is also giving away 50 $200 Devin Max plans — reply with what you're building to enter; winners picked within 24 hours.

Related event: Devin Adds GPT-6 Models, Cutting Task Costs by 61%(2 posts)→

Original post →

More from coding & agent

coding & agent channel →