{"object":"list","data":[{"id":"s1-fast","name":"Fast","description":"Fast open typed-decision profile for text decisions; noul/choice/score answers with probabilities in milliseconds. An alias for the current Fast version (see model_version in every response and the S1-Model-Version header); pin s1-fast-1 or s1-fast-2 to stay on one version. Inputs the current version cannot take (several questions, very long input) are answered by s1-fast-1.","status":"available","modalities":["text"],"question_types":["noul","choice","score"],"max_questions":1,"max_input_tokens":8192,"tier_availability":{"eu":true,"global":true},"prices":{"price_version":"S1-LAUNCH-6","unit":"per_million_input_tokens","output_tokens":"free","launch_pricing":true,"USD":{"global":"0.019","eu":"0.021"},"EUR":{"global":"0.017","eu":"0.018"}},"aliases":["s1-fast-1","s1-fast-2"],"attribution":"s1-fast-2 is based on Blink v0.3 26B-A4B by PixilabAI (Apache-2.0), itself derived from Gemma 4 26B-A4B by Google DeepMind (Apache-2.0); served in NVFP4 by System1 Models. s1-fast-1 is built on Plumb-4B (Apache-2.0)."},{"id":"s1-fast-2","name":"Fast 26B-A4B (NVFP4)","description":"Current Fast version: a 26B-A4B open typed-decision model; noul/choice/score answers with probabilities. Text only, exactly one question per request, at most 8,192 input tokens and 26 options per question. Billed at the s1-fast price.","status":"available","modalities":["text"],"question_types":["noul","choice","score"],"max_questions":1,"max_input_tokens":8192,"tier_availability":{"eu":true,"global":true},"prices":{"price_version":"S1-LAUNCH-6","unit":"per_million_input_tokens","output_tokens":"free","launch_pricing":true,"USD":{"global":"0.019","eu":"0.021"},"EUR":{"global":"0.017","eu":"0.018"}},"version_of":"s1-fast","attribution":"s1-fast-2 is based on Blink v0.3 26B-A4B by PixilabAI (Apache-2.0), itself derived from Gemma 4 26B-A4B by Google DeepMind (Apache-2.0); served in NVFP4 by System1 Models."},{"id":"s1-pro","name":"Pro","description":"Larger open typed-decision profile for text decisions; noul/choice/score answers with probabilities. An alias for the current Pro version: since 12 October 2026 s1-pro-3, based on vLLM-SR d3 26B (Apache-2.0), served in FP8 (see model_version in every response and the S1-Model-Version header). Pin s1-pro-3, or s1-pro-2 (the previous version, Perplexity Decider v1.1, available until at least 11 November 2026), to stay on one version. Text only; up to 3 questions per request, at most 8,192 input tokens per question including the state.","status":"available","modalities":["text"],"question_types":["noul","choice","score"],"max_questions":3,"max_input_tokens":8192,"tier_availability":{"eu":true,"global":true},"prices":{"price_version":"S1-LAUNCH-6","unit":"per_million_input_tokens","output_tokens":"free","launch_pricing":true,"USD":{"global":"0.08","eu":"0.084"},"EUR":{"global":"0.07","eu":"0.074"},"existing_accounts":{"price_version":"S1-LAUNCH-5","unit":"per_million_input_tokens","output_tokens":"free","launch_pricing":true,"USD":{"global":"0.038","eu":"0.04"},"EUR":{"global":"0.033","eu":"0.035"},"until":"2026-11-12T00:00:00Z"}},"aliases":["s1-pro-1","s1-pro-2"],"attribution":"s1-pro-2 is based on Perplexity Decider v1.1 27B (Apache-2.0), itself derived from Qwen3.8-27B by Alibaba Cloud (Apache-2.0); served in NVFP4 by System1 Models. s1-pro-1 is built on Surogate Rune 26B-A4B v3 (Apache-2.0) and Gemma 4 by Google DeepMind; fallback Winnow-12B by EldanRing."},{"id":"s1-pro-2","name":"Pro 27B (NVFP4)","description":"Current Pro version: a 27B open typed-decision model served in NVFP4; noul/choice/score answers with probabilities. Text only; up to 3 questions per request, each answered separately, with at most 8,192 input tokens per question including the state. Billed at the s1-pro price.","status":"available","modalities":["text"],"question_types":["noul","choice","score"],"max_questions":1,"max_input_tokens":8192,"tier_availability":{"eu":true,"global":true},"prices":{"price_version":"S1-LAUNCH-6","unit":"per_million_input_tokens","output_tokens":"free","launch_pricing":true,"USD":{"global":"0.08","eu":"0.084"},"EUR":{"global":"0.07","eu":"0.074"},"existing_accounts":{"price_version":"S1-LAUNCH-5","unit":"per_million_input_tokens","output_tokens":"free","launch_pricing":true,"USD":{"global":"0.038","eu":"0.04"},"EUR":{"global":"0.033","eu":"0.035"},"until":"2026-11-12T00:00:00Z"}},"version_of":"s1-pro","attribution":"s1-pro-2 is based on Perplexity Decider v1.1 27B (Apache-2.0), itself derived from Qwen3.8-27B by Alibaba Cloud (Apache-2.0); served in NVFP4 by System1 Models."},{"id":"quyet-1-large","name":"Quyet-1.0-Large","description":"Named open typed-decision model Quyet-1.0-Large (FP8); noul/choice/score answers with probabilities. Text only, exactly one question per request, at most 8,192 input tokens and 10 options per question. Billed at its own price, see prices.","status":"available","modalities":["text"],"question_types":["noul","choice","score"],"max_questions":1,"max_input_tokens":8192,"tier_availability":{"eu":true,"global":true},"prices":{"price_version":"S1-LAUNCH-6-QUYET","unit":"per_million_input_tokens","output_tokens":"free","launch_pricing":true,"USD":{"global":"0.035","eu":"0.038"},"EUR":{"global":"0.031","eu":"0.033"}},"version_of":"s1-pro","attribution":"Quyet-1.0-Large by Chinh Nguyen (Apache-2.0), derived from Gemma 4 by Google DeepMind (Apache-2.0); served in FP8 by System1 Models. Quyet by Chinh Nguyen."},{"id":"s1-vision","name":"Vision","description":"Image + text typed-decision profile: one image plus state, noul/choice/score answers with probabilities. Requests that contain an image are billed at the image_requests rate. An alias for the current Vision version (see model_version in every response and the S1-Model-Version header); pin s1-vision-1 or s1-vision-2 to stay on one version.","status":"available","modalities":["text","image"],"question_types":["noul","choice","score"],"max_questions":1,"max_input_tokens":8192,"tier_availability":{"eu":true,"global":true},"prices":{"price_version":"S1-LAUNCH-6","unit":"per_million_input_tokens","output_tokens":"free","launch_pricing":true,"USD":{"global":"0.031","eu":"0.032"},"EUR":{"global":"0.027","eu":"0.028"},"image_requests":{"applies_to":"requests_with_image","price_version":"S1-LAUNCH-3-IMAGE","unit":"per_million_input_tokens","output_tokens":"free","launch_pricing":true,"USD":{"global":"0.217","eu":"0.228"},"EUR":{"global":"0.19","eu":"0.2"}}},"attribution":"s1-vision-2 is based on Blink v0.3 26B-A4B by PixilabAI (Apache-2.0), itself derived from Gemma 4 26B-A4B by Google DeepMind (Apache-2.0); served in NVFP4 by System1 Models. s1-vision-1 is built on Surogate Rune 26B-A4B v3 (Apache-2.0), Jev-Omni by akhilaaa3 (Apache-2.0) and Gemma 4 by Google DeepMind. Jev is a trademark of TypeSafe AI; System1 Models is not affiliated with TypeSafe.","aliases":["s1-vision-1","s1-vision-2"]},{"id":"s1-vision-2","name":"Vision 26B-A4B (NVFP4)","description":"Current Vision version: one image plus state, noul/choice/score answers with probabilities. Exactly one question per request, at most 8,192 input tokens and 26 options per question. Requests that contain an image are billed at the image_requests rate.","status":"available","modalities":["text","image"],"question_types":["noul","choice","score"],"max_questions":1,"max_input_tokens":8192,"tier_availability":{"eu":true,"global":true},"prices":{"price_version":"S1-LAUNCH-6","unit":"per_million_input_tokens","output_tokens":"free","launch_pricing":true,"USD":{"global":"0.031","eu":"0.032"},"EUR":{"global":"0.027","eu":"0.028"},"image_requests":{"applies_to":"requests_with_image","price_version":"S1-LAUNCH-3-IMAGE","unit":"per_million_input_tokens","output_tokens":"free","launch_pricing":true,"USD":{"global":"0.217","eu":"0.228"},"EUR":{"global":"0.19","eu":"0.2"}}},"attribution":"s1-vision-2 is based on Blink v0.3 26B-A4B by PixilabAI (Apache-2.0), itself derived from Gemma 4 26B-A4B by Google DeepMind (Apache-2.0); served in NVFP4 by System1 Models.","version_of":"s1-vision"},{"id":"s1-llm-auto-router","name":"s1-llm-auto-router (mmBERT-small, 11 layers, CPU)","description":"Fixed-schema routing classifier for LLM auto-routers: answers category, difficulty, stakes and four routing flags for one request in a single CPU pass (~26 ms). Not a general typed-decision model.","status":"available","modalities":["text"],"question_types":["noul","choice","score"],"fixed_questions":{"category":"choice","difficulty":"score","stakes":"score","needs_tools":"noul","needs_vision":"noul","needs_long_context":"noul","follow_up":"noul"},"max_questions":7,"tier_availability":{"eu":true,"global":true},"prices":{"price_version":"S1-LAUNCH-6","unit":"per_million_input_tokens","output_tokens":"free","launch_pricing":true,"USD":{"global":"0.001","eu":"0.001"},"EUR":{"global":"0.001","eu":"0.001"}},"attribution":"Fine-tuned from jhu-clsp/mmBERT-small (MIT) by System1 Models on synthetic routing requests labelled by an LLM ensemble."},{"id":"s1-pro-3","name":"Pro d3 26B","description":"Current Pro version: vLLM-SR d3 26B, an open typed-decision model (Apache-2.0, by the vLLM Semantic Router project), served in FP8; noul/choice/score answers with probabilities. Text only; up to 3 questions per request, each answered separately, with at most 8,192 input tokens per question including the state. Billed at the s1-pro price.","status":"available","modalities":["text"],"question_types":["noul","choice","score"],"max_questions":3,"max_input_tokens":8192,"tier_availability":{"eu":true,"global":true},"prices":{"price_version":"S1-LAUNCH-6","unit":"per_million_input_tokens","output_tokens":"free","launch_pricing":true,"USD":{"global":"0.08","eu":"0.084"},"EUR":{"global":"0.07","eu":"0.074"},"existing_accounts":{"price_version":"S1-LAUNCH-5","unit":"per_million_input_tokens","output_tokens":"free","launch_pricing":true,"USD":{"global":"0.038","eu":"0.04"},"EUR":{"global":"0.033","eu":"0.035"},"until":"2026-11-12T00:00:00Z"}},"version_of":"s1-pro","attribution":"vLLM-SR d3 (Apache-2.0), based on Qwen3.8-27B by Alibaba Cloud."}],"routing_policy":"Only models that have passed their launch gates and are currently available are listed. Concrete model IDs are not automatically rerouted."}