RightNow, a Y Combinator Fall 2026 startup, launched its first “System Zero” model on 30 October. It takes in text, JSON, robot state, code, PDFs, images, voice and video and returns a decision with a probability for each option while the input is still arriving. Its price page gives an example call at USD 0.00028; a million such calls would cost USD 280, by our arithmetic.
The company posted at 23:28 UTC: “RightNow is out today. our first โSystem Zeroโ model”. It says it makes “30 ms decisions on text” and handles “up to 1,024 questions in one call”.
Short explainer · RightNow Prices AI Decisions at 0.00028 Dollars a Call
A vertical news Short narrated from this piece’s headline and opening line.
Business Pill · AI AGENTS
A short explainer on what AI agents are and how they work. It teaches the general idea only and says nothing about RightNow.
The key insight: RightNow charges per answered question and per call on top of the tokens it reads, USD 0.00028 for its 500-token, one-question example, or USD 280 per million such calls by our arithmetic; its 63.20 decision score is self-measured, not an Artificial Analysis evaluation.
RightNow’s homepage lists the inputs as “text, JSON, robot state, codebase, PDF, images, voice, video (preview)”. The standard context is 38,000 tokens, and the site says a “1M context window” can be turned on in the console settings.
Jaber, a founder of RightNow, wrote on X that most decision models “read text, wait until the whole input has arrived, and then give you an answer”. He added that this “doesnโt work for a robot, a phone call, or a video feed, where the input never really ends”.
He wrote that the team “spent half of our YC money training a model” and that the model “will stay closed source and available via RightNow API”. He also called it “our first checkpoint” and said it is “still a little buggy”.
Decision-model scores as shown on RightNow’s homepage. Scores for other models are Artificial Analysis Decision Index values published by Artificial Analysis; RightNow 3.0.0’s 63.20 is RightNow’s own measurement on the public held-out half (28 October), not an Artificial Analysis evaluation, and RightNow says the order is indicative. The 1.8-point gap is our arithmetic.
What It Costs
The price page lists uncached text at USD 0.19 per million tokens, USD 0.00013 per answered question and USD 0.000055 per answered call. Its worked example, a 500-token call with one answered question, comes to USD 0.00028; new accounts get USD 5 free at signup.
Of the three charges in that example, the per-question fee is the largest, by our arithmetic. The company’s launch post also says “output is free”.
Where It Says It Ranks
RightNow puts its own model, RightNow 3.0.0, at 63.20 on the Artificial Analysis Decision Index. A footnote says the score was “measured by RightNow AI on the public held-out half, 2026-10-28. Not an Artificial Analysis evaluation”.
The same table shows scores published by Artificial Analysis for other models: 65 for Liquid AI’s d1-ultra, Subquadratic’s SubQ Snap 1.1 and Baseten’s Elephant 9B, 56 for Perplexity’s Decider v1.2, 54 for Microsoft-Decision-1 and 48 for OpenAI’s GPT-6 Luna (Decisions). RightNow notes the order is “indicative, not an Artificial Analysis ranking”.
The Business Lever: Pricing a Decision, Not a Conversation
As we read the price page, RightNow charges for each answered question and each call on top of the tokens it reads, so a customer pays per decision taken. Jaber said the company will publish a technical paper on what it calls Reflex Training and Reflex Architecture.
The Structural Read
As we read RightNow’s price page, the per-question fee is the largest of the three charges in its worked example, so cost scales with the number of decisions asked for.
As we read the founder’s post, the company is aiming at inputs that never finish, such as a robot, a phone call or a video feed, where a request-and-wait model has to stop and answer.
As we read the benchmark table, RightNow places its own score among models Artificial Analysis evaluated, while flagging that its row comes from a different evaluation and that the order is indicative.
RightNow on X, 30 October 2026
“RightNow is out today. our first โSystem Zeroโ model”
Three Implications
FOR DEVELOPERS RightNow lists USD 0.19 per million uncached text tokens and gives USD 5 free at signup.
FOR RIVAL DECISION MODELS The table RightNow shows lists scores of 65 for d1-ultra, SubQ Snap 1.1 and Elephant 9B.
FOR WHAT IS STILL UNKNOWN RightNow lists its cost per million tasks as pending, and its 63.20 is not an Artificial Analysis evaluation.
The framework argues: “Inference is a revenue engine: continuous, distributed across millions of applications and agents, incurring marginal cost with every query.”
As we read it, a model priced per answered question is one way to charge for that marginal cost. The framework predates RightNow’s launch and does not discuss it.
What Is Not Established
RightNow’s 63.20 is its own measurement, not an Artificial Analysis evaluation, and the site lists its cost per million tasks as pending, so how it compares on cost with the models Artificial Analysis scored is not established. The 30 ms figure is the company’s claim; we did not test the model, and customer names and funding beyond Y Combinator are not established in what we read.
This Week In The AI Supercycle
Execution Fault Lines, Concentrated Demand, and Geopolitical Firewalls
This week’s issue: the AI build is not short of demand for its parts. It is short of cheap money, and its end revenue rests on a narrow base of large, often venture-funded buyers.
RightNow launched a closed-source model that streams in text, code, voice, video and robot state and returns per-option probabilities as data arrives. Its price page gives USD 0.00028 for a 500-token, one-question call, or USD 280 per million such calls by our arithmetic; its 63.20 decision score is self-measured.
A note on sourcing. Prices, inputs, context limits and benchmark figures are from RightNow’s homepage and docs, read on 31 October; the 63.20 score is RightNow’s own measurement, not an Artificial Analysis evaluation. Quotes are from posts on X by RightNow and its founder. We did not test the model or contact the company. Figures labelled our arithmetic are our own.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
