Skip to content
The RightNow mark over a green hill and a blue sky.

RightNow System Zero

RightNow is our first decision model. It belongs to a category we call System Zero.

It doesn't write text. You send it a state and your questions, and for each question it returns a decision with a probability for every option, a confidence, and a calibration version. Output is not charged, and refused calls are not charged.

RightNow AI is a YC-backed GPU research lab. This is the first checkpoint of our post-trained models, and more checkpoints will follow.

The step before thinking

Reflex Training. Decision models usually read the whole input, then answer. RightNow decides on whatever has arrived so far: stream in text, JSON, images, voice, PDF and video, and it returns a probability for every option at any moment. It is trained to hold back until the evidence is there, then commit. That is the step before thinking.

Reflex Architecture. System Zero keeps a probability for every option and updates it as each part of the input arrives.

One model, eight inputs

RightNow takes text, JSON, robot state, codebases, PDF, images, voice and video (preview).

Text
JSON
Robot state
Codebase
PDF
Images
Voice
Video (preview)preview

Supported formats are PNG and JPEG for images; WAV, FLAC, MP3, OGG (Vorbis, Opus) and MP4 audio at 16 kHz; and PDF for documents.

Each call takes:

  • up to 1,024 questions
  • up to 38,000 state tokens
  • up to 16 attachments, including up to 4 images, 1 audio file and 1 video

Text questions take 2 to 10 options or 2 to 4 levels. Media questions take 2 to 26 options or 2 to 10 levels.

Streaming decisions

You don't have to wait for the full input. Send changes as they happen: append text, update the state, or add attachments. RightNow returns versioned decisions. Every new decision carries its version, and the probability for every option updates as each part of the input arrives.

An example

State
The blender I ordered arrived with a cracked jar.
Question (boolean)
Did the item arrive damaged?
Decision
true
truetrue: 0.9995010.999501
falsefalse: 0.0004990.000499
Recorded example, system-zero-2.0.0, Oct 26, 2026.

Speed

Decision latencyIncluding the input
0ms506
Text
Text: 30 to 33 ms, p95 32 to 34 ms30 to 33 ms
Session
Session: 34 to 36 ms34 to 36 ms
Image
Image: 41 ms41 msIncluding the input: 178 ms178 ms
Audio
Audio: 34 ms34 msIncluding the input: 96 ms96 ms
Video (preview)
Video (preview): 41 ms41 msIncluding the input: 168 ms168 ms
Document
Document: 46 ms46 msIncluding the input: 506 ms506 ms
MODEL-SPEED-ROW-1027
Medians. Record MODEL-SPEED-ROW-1027, Oct 27, 2026.
InputDecision latencyIncluding the input
TextDecision latency30 to 33 ms (p95 32 to 34 ms)Including the input
Single tokenDecision latency7.2 msIncluding the input
SessionDecision latency34 to 36 msIncluding the input
ImageDecision latency41 msIncluding the input178 ms
AudioDecision latency34 msIncluding the input96 ms
Video (preview)Decision latency41 msIncluding the input168 ms
DocumentDecision latency46 msIncluding the input506 ms

Medians. Record MODEL-SPEED-ROW-1027, Oct 27, 2026.

Benchmarks

Artificial Analysis Decision Index

RightNow 3.0.0 scores 63.20. This was measured by RightNow AI on the public held-out half. It is not an Artificial Analysis evaluation. The published rows come from Artificial Analysis's own evaluation, so the two are different evaluations. The chart shows positions and values only.

Index0100
lib/system-zero-aa-decision.generated.json
RightNow 3.0.0: measured by RightNow AI on the public held-out half, 2026-10-28. Not an Artificial Analysis evaluation. Published rows: Artificial Analysis (artificialanalysis.ai), downloaded Oct 30, 2026.

Amazon Polarity

Accuracy
97.33%
Cost per case
USD 0.0005726
Time per case
0.3371 s
Accuracy: 97.33%. Cost per case: USD 0.0005726. Time per case: 0.3371 s. Amazon Polarity. Run BENCH-FRONTIER-1026A, Oct 26, 2026.

Invoice Processing and Security Incidents

ModelAgreementUSD per workflowSeconds per workflow
RightNowAgreementPendingUSD per workflowPendingSeconds per workflowPending
Claude Haiku 4.5Agreement42.9%USD per workflow$0.0558Seconds per workflow30.8 s
Claude Opus 5Agreement78.4%USD per workflow$0.4856Seconds per workflow92.1 s
Claude Sonnet 5Agreement72.9%USD per workflow$0.3616Seconds per workflow241.3 s
DeepSeek V4 FlashAgreement69.8%USD per workflow$0.0133Seconds per workflow84.1 s
DeepSeek V4 ProAgreement72.7%USD per workflow$0.0830Seconds per workflow137.4 s
GPT-5.6 LunaAgreement67.8%USD per workflow$0.0081Seconds per workflow21.4 s
GPT-5.6 SolAgreement79.1%USD per workflow$0.2152Seconds per workflow34.3 s
GPT-5.6 TerraAgreement74.7%USD per workflow$0.0778Seconds per workflow17.3 s
System One 1.13.0Agreement61.8%USD per workflow$0.0011Seconds per workflow0.5 s

System One's published workflow boards, snapshot Sept 28, 2026. RightNow: Pending.

RightNow and System One

CapabilityRightNowSystem One
InputsRightNowtext, JSON, robot state, codebase, PDF, images, voice, video (preview)System OneText, JSON; images opt in.
VoiceRightNowWAV audioSystem OneNot supported
Video (preview)RightNowMP4 video (preview)System OneNot supported
FilesRightNowPNG, JPEG, WAV, PDF, MP4System OnePNG, JPEG (opt in)
Standard contextRightNow38,000 tokensSystem One64k per-request / 32k per-question
Questions per callRightNowMultiple questions per callSystem OneMultiple questions per call
ProbabilitiesRightNowPer-option probabilitiesSystem OneTyped probabilities
StreamingRightNowStreamingSystem One
versioned decisionsRightNowSystem One

System One column from its docs, read Oct 28, 2026.

Pricing

ItemPrice
Text inputPrice0.19 per million tokens
Cached text inputPrice0.019 per million tokens
OutputPriceNot charged
Answered questionPrice0.00013 per answered question
Call feePrice0.000055 per answered call
ImagePrice0.011 per image
AudioPrice0.0220000002 per minute
PDFPrice0.011 per page
Video (preview)Price0.0018 per second

USD. Rate card sz-card-2026-10-27-g, as of Oct 27, 2026.

The call fee is never charged on refusals, zero-answer calls, or free, promotional or replay calls.

Two reference costs:

  • 356 uncached tokens with one answered question costs USD 0.00025
  • 38,000 tokens with one question costs USD 0.0074

New accounts get USD 5 in credit at sign-up. It's one per eligible verified email, subject to eligibility checks, and valid for 12 months. Top-ups start at USD 5.

Start building