Frontier Log

Cognition

SWE-2 Flash

Released · cognition.com

Price

Per 1M tokensInputCachedOutput
Rate for all three effort variants in Devin Desktop and CLI; free on self-serve plans until 2026-12-04 18:00 UTC; cache writes $0.125$0.10$0.01$0.50
Price history
DateInputCachedOutput
As of docs.devin.ai · Checked$0.10$0.01$0.50

Benchmark results

BenchmarkScore
Terminal-Bench 4.0harness not stated20.0%Checked A person checked this number against its source on 7 Nov 2026. This confirms the number matches what the lab published. It doesn't confirm that the score was reproduced.
FrontierCode 1.1Main set · harness not stated45.5%Checked A person checked this number against its source on 7 Nov 2026. This confirms the number matches what the lab published. It doesn't confirm that the score was reproduced.
DeepSWE v1.1harness not stated60.1%Checked A person checked this number against its source on 7 Nov 2026. This confirms the number matches what the lab published. It doesn't confirm that the score was reproduced.
Terminal-Bench 2.1harness not stated89.2%Checked A person checked this number against its source on 7 Nov 2026. This confirms the number matches what the lab published. It doesn't confirm that the score was reproduced.
Sources

Reported by Cognition

Related

All models