Clef (Cloudflare)https://blog.cloudflare.com/clef-decision-models/Cloudflare's open decision models Clef and Clef-flash (Clef-omni adds audio and video), hosted on Workers AI and released as Apache-2.0 weights. Drop-in compatible with Jev's System One API; Cloudflare reports Clef 2.5x and Clef-flash 13x faster than Jev at the median. Clef-flash costs $0.038 per 1M input tokens. d1 (Liquid AI)https://www.liquid.ai/blog/d1-decision-modelLiquid AI's decision-model family: hosted d1.1 and d1-ultra (text and images) over the Liquid API, plus open-weight d1-3B and d1-omni-600M (text with image or audio) for local and edge use under the LFM Open License. Uses Jev's request shape. Decisions API (OpenAI)https://openai.com/index/decisions-api/OpenAI's Decisions API, generally available 22 October 2026 on GPT-6 Luna: predicate, choice and score questions over text, images and JSON, up to 400 questions per request, $0.10 per 1M input tokens ($0.01 cached). Its own request shape, not Jev's. GLiDE (Fastino)https://fastino.ai/blog/decision-models-explainedFastino's Generalized Lightweight Decision Engine: a hosted API using the same state-and-questions format as Jev, which reasons further only when its first estimate is uncertain. Fastino's open GLiNER2.5-Decide is a compact English classifier under Apache-2.0. Jev (TypeSafe AI)https://typesafe.aiThe model that defined the category: TypeSafe's hosted System One decision model. One POST to /v1/systemone with a state and typed questions returns calibrated probabilities. Jev 1.14 (jev-1.14.0) has been available since 14 October 2026; text input, closed weights. Kev (Jared Palmer)https://github.com/jaredpalmer/kevA family of small self-hosted decision models built on Qwen3.5 (0.8B, 4B and 9B) with training code and an API that matches TypeSafe's System One, so TypeSafe's own SDK can point at a local server. Apache-2.0. Laya (Convai Innovations)https://github.com/NandhaKishorM/layaOpen, non-autoregressive decision engine: typed choice, score and yes/no decisions in a single forward pass, 100+ languages, with a router that picks the checkpoint per request. Apache-2.0; runs on your own hardware (33 ms for one question on a T4). Microsoft-Decision-1https://commandline.microsoft.com/microsoft-decision-1-model-foundry/Microsoft's decision-scoring model, post-trained from Qwen3.5-9B, in public preview in Microsoft Foundry and on OpenRouter. $0.042 per 1M input tokens, matching Jev; Microsoft says it will be rebased on MAI and OpenAI models. pplx-decider (Perplexity)https://docs.perplexity.ai/docs/decisions/quickstartPerplexity's Decisions API: pplx-decider-v1.2 (27B, $0.02 per 1M input tokens) and pplx-decider-flash-v1.2 (4B, $0.01). Text and image input, 256K context; the flash model also runs locally in Decider for Mac.