Artificial intelligence news.
Straight from the original announcements.

Hippocratic AI Launches Polaris 6 Ultra, a Healthcare Voice Model With 'Bedside Intuition'

November 2, 2026, 22:41By the AI News Bites Editorial TeamSource: PR Newswire
Hippocratic AI Launches Polaris 6 Ultra, a Healthcare Voice Model With 'Bedside Intuition'
Image: PR Newswire

Hippocratic AI has announced Polaris 6 Ultra, a new generation of its healthcare language model that, according to the company, can read a patient's emotional state and plan its next sentence while still speaking. The company describes this as "bedside intuition" and says the release brings six upgrades to how an AI care agent listens, thinks and speaks.

Key points
  • Hippocratic AI announced Polaris 6 Ultra, which plans its next sentence while the previous one is still being spoken.
  • The safety constellation grew to 7.4 trillion parameters, with 30 specialist supervisors and 142 production monitors.
  • The company reports a 1.3x end-to-end speedup over Polaris 5 and a 0.28 second time to first sentence.
  • Clinical ASR cut misheard medications or symptoms from 1.19 to 0.20 per 100 calls on 2,778 reviewed calls.
  • Polaris 6.0 comes in Ultra, Pro, Flash and Nano tiers, with new flexible pricing.

Planning the next sentence while speaking

The central feature is what Hippocratic AI calls AI Intuition (v1). According to the company, the model generates an unspoken planning span before every sentence while the audio of the previous sentence is still playing. In that span it reads the patient's mood and resistance and decides what to raise now and what to save for later.

Time to first sentence stays at 278 milliseconds for the LLM, the company says. Each plan is logged against the sentence it produced and shown in the Safety Portal beside the transcript, so a clinician can audit, sentence by sentence, which risk the model noticed and which it deferred.

The voice has also been updated. Care Voice (v3) is described as the first Polaris voice directed by the model rather than narrated. Breath, sighs, laughter, pauses and tempo are emitted as tokens from the plan, so pacing adapts to the patient's state. Hippocratic AI says its in-house text-to-speech model pronounces 90% of 629 evaluated drug names correctly, versus 75% for the leading commercial TTS.

Better hearing, more common sense and a larger safety constellation

Common Sense 2.0 applies full-parameter reinforcement learning to the 671B primary model, end to end for the first time in a Polaris release. According to the company, this moved its internal healthcare-intelligence benchmark 18 points over its own base and 11 points past the best frontier model. The largest gains came in escalation and transfer, the first thirty seconds of a call, language switching and call endings.

Clinical ASR (CASR v3) is an in-house audio-LLM conditioned on the conversation and the chart. Hippocratic AI says it hears patients who mumble, are recovering from a stroke, or are calling from a landline with the television on. On 2,778 clinician-reviewed production calls, the company reports:

  • Misheard medications or symptoms fell from 1.19 to 0.20 per 100 calls.
  • Misheard numbers, names and emails fell from 0.57 to 0.00.
  • Words invented on noise-only audio fell from 24% to 0.6%.
  • Television speech attributed to the patient fell from 64.8% to under 3%.

The 671B primary model now runs inside a constellation of 30 specialist supervisors and 142 production monitors, totaling 7.4 trillion parameters, up from 5.0 trillion in Polaris 5. They check every turn in parallel across areas such as escalation and crisis response, medications, identity and consent, scheduling, privacy and conversational quality.

Speed and the model lineup

Hippocratic AI reports a 1.3x end-to-end speedup over Polaris 5 as patients experienced it. It credits Blackwell-native 4-bit numerics, routing-aware mixture-of-experts scheduling, first-token speculative decoding and a serving engine the company owns. On byte-identical prompts on a 671B parameter LLM, the company says its stack reaches the first sentence in 0.28 seconds, versus 0.71 to 0.94 seconds for the four fastest commercial providers.

Polaris 6.0 ships in four tiers so customers can match coverage and speed to their needs:

  • Ultra: the full constellation with all 31 supervisors, for clinical patient-facing conversations.
  • Pro: sub-clinical workloads.
  • Flash: high-volume, subclinical use.
  • Nano: administrative use cases.

The company also points to increased EHR integration and new flexible pricing, though the announcement does not give price figures.

What it means for healthcare buyers

CEO Munjal Shah said Polaris 6.0 was built from more than 323 million patient conversations. Chief Science Officer Subhabrata Mukherjee said the team refused the usual trade-off in healthcare voice AI, where speech-to-speech models are "fast and shallow" and reasoning models are "capable and slow."

On safety, Hippocratic AI says the clinician-measured safety of the Polaris constellation has held at its ceiling since Polaris 4, with a minor-harm rate in clinician output testing of 0.00% for the latest constellation. It says Polaris 6 Ultra ships under the same criterion as every release: clinically tested, validated by licensed clinicians and governed under human oversight before it reaches a patient.

The company says its agents have completed more than 250 million patient voice interactions across more than 300 live clinical use cases, with zero safety issues in production. Its agents do not diagnose or prescribe. All performance figures in the announcement come from Hippocratic AI's own testing and have not been independently verified in the source.

Hippocratic AIPolaris 6 UltraPolaris 6.0MENLO PARKlanguage modelshealthcarevoice and speechenterprise AI

Related news

Snorkel AI says gated training lifts Qwen3.8-27B on professional tasks
ResearchNovember 19, 2026, 15:32

Snorkel AI says gated training lifts Qwen3.8-27B on professional tasks

Snorkel AI says it raised Qwen3.8-27B's pass@1 performance on OpenAI's GDPval gold set from 31.9% to 40.7% by combining expert-built training data with binary-gated…

Business
BusinessNovember 19, 2026, 10:00

Youdao reports lower revenue but ninth straight quarter of operating profit

Youdao reported RMB1.4 billion in third quarter 2026 net revenue, down 12.3% from a year earlier, while operating income rose 19.2% to RMB33.7 million. The company said…

AWS brings full fine-tuning to SageMaker AI’s serverless customization
Products and toolsNovember 19, 2026, 01:15

AWS brings full fine-tuning to SageMaker AI’s serverless customization

AWS has added Full Fine-Tuning (FFT) support to SageMaker AI Serverless Model Customization, allowing businesses to update all model parameters without provisioning GPUs…