Articul8 launches Bivia, a multimodal 'decision layer' model that outputs zero generated tokens
testingcatalog · x · 2026-10-29
AI company Articul8 introduced Bivia, pitched as "the decision layer for your AI" — a compact multimodal model that reads images, tables, charts, documents, signals, audio and video directly as state and returns structured decisions (choice, boolean, score, span, extraction) with zero generated tokens.
Key points:
- Calibrated by design: every answer ships with a real confidence score; an abstain mode hands uncertain calls back to humans
- Built for agents: targets the inefficiency of calling a giant LLM for tiny uncertain decisions ("yes / option B / 3.7"), which drives latency and token bills across thousands of calls per day
- 9 decision types, 7 input modalities, 3 operation modes (Fast/Deep/Thinking), 50+ languages
- Launch offer: first 10,000 users get 1 billion free tokens
Articul8 also develops domain-specific models for energy, manufacturing, and aerospace.
Related event: Articul8 Launches Bivia, a Multimodal AI Decision Model(3 posts)→
More from coding & agent
- Lean 4-person AI startup Physera hires to build training and eval infra for multimodal coding agents — himanshustwts · 2026-10-29
- Using a Claude subscription inside third-party apps runs into ToS limits — ZookeepergameOdd6078 · 2026-10-29
- Matt Pocock fixes Codex subagents inheriting parent context instead of starting fresh — mattpocockuk · 2026-10-29
- Indie dev builds an LLM cost guard that alerts on unusual API usage — WittyBoysenberry9860 · 2026-10-29
- Claude guessed LLMs won't write full PRs until 2028 — even models can't grasp their own pace — felixrieseberg · 2026-10-29
- safeagents adds --log-only mode: observe agent calls without blocking, with live dashboard — yb2698 · 2026-10-29