Coding models are now priced like components. cognition's swe-2, out september 13, is a cheaper…
coding models are now priced like components. cognition's swe-2, out september 13, is a cheaper agentic coding model shipped straight inside devin, making the autonomous agent's brain a replaceable part.
the model is becoming an input to the product, not the product.
Context
Cognition's SWE-2 post, dated 10 September 2026, reports 50.0 percent on FrontierCode 1.1 Main, within one point of Fable 5.1 while 64 percent cheaper, a model post-trained from Kimi K3 and available from that day in Devin Desktop and the CLI, with a rollout to Devin Web and Fusion. Costs in its charts assume list pricing including public discounts. The same table shows Terminal-Bench 4 at 27.3 percent for SWE-2 against 55.8 percent for Fable 5.1 and 57.9 percent for GPT-6 Astra. Devin's current pricing page, read on 4 October 2026, lists Free, Pro at 20 dollars a month, Max at 200 dollars a month and a Teams plan, and Pro lists free use of SWE-2 Free in Devin Desktop and CLI through 10 October 2026.
The first-party date is 10 September and not 13 September. The 64 percent is a vendor ratio against one model on one benchmark using list prices and the vendor's harness, and SWE-2 does not match Fable 5.1 on Terminal-Bench 4 in the same table. The Devin tiers are a later window than the note. A per-token SWE-2 price was not found in first-party text; one secondary page lists 3 and 15 dollars per million tokens for input and output and that is unverified. Shipped inside Devin is supported for Desktop and CLI. That the model is becoming an input to the product is the author's opinion.
Related work
- Earlier note on Cognition's funding and post-training ↗Cognition context for the same model line.
Watch next
- An independent SWE-2 benchmark run and a first-party per-token price.
Sources
- SWE-2 (Cognition, 10 Sep 2026)cognition.com
- Devin pricingdevin.ai
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 20 September 2026 at 17:45 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →