{"version":1,"type":"story","url":"https://digestai.news/story/cognition-says-swe-2-flash-runs-three-times-faster-and-costs-one-fifth","json":"https://digestai.news/story/cognition-says-swe-2-flash-runs-three-times-faster-and-costs-one-fifth.json","markdown":"https://digestai.news/story/cognition-says-swe-2-flash-runs-three-times-faster-and-costs-one-fifth.md","slug":"cognition-says-swe-2-flash-runs-three-times-faster-and-costs-one-fifth","headline":"Cognition says SWE-2 Flash runs three times faster and costs one-fifth","summary":"Cognition announced SWE-2 Flash on November 4, 2026, saying the coding model runs at three times the speed and one‑fifth the price of its larger SWE-2 sibling. The model, built on the Inkling‑Small mixture‑of‑experts base from Thinking Machines Lab, has 276 billion total parameters and 12 billion active parameters. It is offered through Devin Desktop and CLI and is free to use through November 30, 2026.\n\nCognition reports mixed benchmark results: Flash beats SWE-2 on FrontierCode 2.0 but trails on Terminal‑Bench 4.0 and DeepSWE 1.1. The company attributes most gains to higher‑quality, diverse training data and a novel mid‑training stage before reinforcement learning. Weights are not publicly released; instead Cognition is creating a secure sharing program for trusted industry partners and research labs, citing security concerns after internal testing raised exploit scores from 11.3% to 34.8%.","keyPoints":["Cognition reports SWE-2 Flash runs at three times the speed and costs one-fifth of SWE-2","Flash has 276 billion total parameters, 12 billion active, and is free until November 30 2026","Weights are withheld; Cognition plans a secure sharing program for trusted partners"],"whyItMatters":"A faster, cheaper coding model could lower development costs and speed up routine software‑agent tasks, shaping how companies adopt AI‑assisted engineering tools.","category":{"slug":"models","name":"Generative AI & Models","url":"https://digestai.news/category/models"},"entities":{"companies":["Cognition","Thinking Machines Lab"],"models":["SWE-2 Flash","Inkling-Small","SWE-2"],"people":[]},"firstPublishedAt":"2026-11-04T20:18:02Z","updatedAt":"2026-11-04T20:18:02Z","sourceCount":1,"hasPrimarySource":false,"sources":[{"outlet":"Unite.AI","title":"Cognition Launches SWE-2 Flash for Faster, Lower-Cost Coding Agents","url":"https://unite.ai/cognition-launches-swe-2-flash-for-faster-lower-cost-coding-agents","publishedAt":"2026-11-04T20:18:02Z","type":"press","primary":false,"lead":true}],"sourceNotes":null,"discussions":[],"thread":null,"cite":{"text":"Digest AI, \"Cognition says SWE-2 Flash runs three times faster and costs one-fifth\", 4 November 2026, https://digestai.news/story/cognition-says-swe-2-flash-runs-three-times-faster-and-costs-one-fifth","publisher":"Digest AI","title":"Cognition says SWE-2 Flash runs three times faster and costs one-fifth","datePublished":"2026-11-04T20:18:02Z","url":"https://digestai.news/story/cognition-says-swe-2-flash-runs-three-times-faster-and-costs-one-fifth"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}