{"rewrite":{"id":"r_787a25438e07672f714ba0d8","clusterId":"c_e48b37a64e51904e88d330c5","slug":"subquadratic-announces-subq-snap-a-decision-model-with-384-000-token-input","model":"deepseek-v4-1-flash","headline":"Subquadratic Announces SubQ Snap, A Decision Model With 384,000-Token Input","summary":"Subquadratic announced SubQ Snap, a decision-making model for agent workflows that handles up to 384,000 tokens of text and images, a context window the company describes as twelve times that of Jev. It processes requests of up to 100,000 tokens in under one second, scores four points higher than Jev on Decision Index version 0.2.1, and costs $0.05 per million input tokens with no charge for output.","whyItMatters":"The pitch targets the narrow decisions agents make constantly, where Subquadratic says current decision models stall at 32,000 tokens, so a wide-context model at a low per-token price aims at workflow cost and latency rather than at the headline reasoning tasks LLMs already handle.","webCardHtml":"\u003cp\u003eSubquadratic built SubQ Snap for the small judgments inside agent workflows: which files are relevant, which tool to use next, whether test results pass, and which steps need a state-of-the-art model. The company says existing decision models are stuck at 32,000 tokens, so they judge without seeing most of the material.\u003c/p\u003e\u003cp\u003eSubQ Snap uses SSA, an attention mechanism Subquadratic developed for long-context processing. It handles images as well as text, and the company reports it beats Jev across most of the Decision Index items, with ACOS, API-Bank and When2Call the exceptions. A public preview version is available at the time of writing.\u003c/p\u003e","blueskyPost":"Subquadratic prices SubQ Snap at $0.05 per million input tokens with output free, which makes the 384,000-token window a question of what agent workflows can afford to read, not what they can technically hold.","twitterPost":"SubQ Snap's 384,000-token window only matters if agents can afford to fill it. Subquadratic prices input at $0.05 per million and output free.","threadsPost":"Subquadratic built SubQ Snap to hold 384,000 tokens, then priced input at $0.05 per million and output at nothing. The window sets the ceiling; the pricing decides whether agent workflows ever fill it. Long context is a billing question as much as an engineering one.","newsletterBlurb":"Subquadratic announced SubQ Snap, a decision-making model for agent workflows that takes up to 384,000 tokens of text and images. The company says it scores four points above Jev on Decision Index 0.2.1 and clears 100,000-token requests in under a second. Input costs $0.05 per million tokens, with output free.","attributionJson":"[{\"source\":\"GIGAZINE\",\"url\":\"https://gigazine.net/news/20261013-subq-snap/\",\"title\":\"Decision-making model \\\"SubQ Snap\\\" with support for 384,000-token input arrives, outperforms Jev on both long and short context windows and also handles images\"}]","lintFlagsJson":null,"lintHits":0,"costUsd":0,"inputTokens":4643,"outputTokens":669,"status":"published","repairAttempts":0,"nextRepairAt":null,"factsAttemptedAt":1791870735,"createdAt":"2026-10-13T05:48:10.000Z","publishedAt":"2026-10-13T05:51:10.000Z","updatedAt":"2026-10-13T05:51:10.000Z"},"cluster":{"id":"c_e48b37a64e51904e88d330c5","canonicalTitle":"38万4000トークンの入力に対応した意思決定モデル「SubQ Snap」が登場、コンテキストウィンドウが長くても短くてもJevより優秀で画像にも対応","representativeArticleId":"a_e1be5f8c8fa8b060d5d28683","sourceCount":1,"writtenSourceCount":1,"writeAttempts":0,"isSolo":true,"entitiesJson":"{\"anime_titles\":[],\"manga_titles\":[],\"work_titles\":[\"SubQ Snap\"],\"studios\":[\"Subquadratic\"],\"people\":[],\"type\":\"announcement\",\"domain\":\"other\",\"is_roundup\":false}","contentType":"news","status":"published","firstSeenAt":"2026-10-13T04:39:00.000Z","lastSeenAt":"2026-10-13T04:39:00.000Z","updatedAt":"2026-10-13T05:51:06.000Z"},"attribution":[{"source":"GIGAZINE","url":"https://gigazine.net/news/20261013-subq-snap/","title":"38万4000トークンの入力に対応した意思決定モデル「SubQ Snap」が登場、コンテキストウィンドウが長くても短くてもJevより優秀で画像にも対応"}],"entities":{"anime_titles":[],"manga_titles":[],"work_titles":["SubQ Snap"],"studios":["Subquadratic"],"people":[],"type":"announcement","domain":"other","is_roundup":false},"keyFacts":null}