Rankings/Deepgram
D

Deepgram

AI NativePrivateUpdated November 15, 2026
Try it free →
Founded
2015
CEO
Scott Stephenson
HQ
San Francisco, CA, United States
Stage
Series C
Revenue
Not disclosed
Total Funding
$229M
Employees
201-500
Flag a data issue
Foundation Models
45.7Bluescore
Momentum—
Market MapDeep Tech
Rank change (7d)—

Executive Summary

Updated Nov 12, 2026

What they do

  • Real-time speech-to-text, text-to-speech and voice agent APIs for developers and enterprises
  • Builds its own speech models, including Nova-3, Flux conversational STT and Flux TTS
  • Is acquiring Daily, maker of the open-source Pipecat framework, for an end-to-end voice stack

Why it matters

  • Raised a $130M Series C at a $1.3B valuation in January 2026, led by AVP
  • Says 5,000+ organizations use it and 1,000+ have built on its Voice Agent API
  • Comparisons often rank it first for streaming latency and speed in voice agent use

Best fit

  • Contact center and CX platforms embedding real-time transcription and voice agents
  • ISVs building voice agents that need low latency and good handling of interruptions
  • Enterprises needing dedicated, EU-hosted or self-managed speech deployments with HIPAA support

Watch-outs

  • Some benchmarks show AssemblyAI and ElevenLabs with lower word error rates than Nova-3
  • The Daily/Pipecat integration and the new Voice Agent Platform are still early
  • Usage-based pricing changes by product and model, so forecast costs at your volume

Products & Offerings

Voice Agent API

A single real-time API that combines speech recognition, LLM orchestration and text-to-speech for building conversational voice agents, now including the Think-Live turn-taking model.

APIDeveloperFlagship2025

Nova-3

Deepgram's main speech-to-text model for streaming and batch transcription, with expanded European and Asian language support.

ModelEnterpriseCore Platform2025

Flux

A conversational speech recognition model built for voice agents that handles interruptions and turn detection without adding latency; a multilingual version became generally available in April 2026.

ModelDeveloperHigh Growth2025

Aura-2 and Flux TTS

Low-latency text-to-speech models offering 50+ natural voices for enterprise voice agents; Flux TTS became generally available in October 2026.

ModelEnterpriseLive2025

Pipecat and Daily real-time infrastructure

The open-source Pipecat voice agent framework and Daily's WebRTC and telephony network, coming through the pending Daily acquisition, which will form the base of Deepgram's upcoming Voice Agent Platform.

InfrastructureDeveloperEmerging

Latest News

Analyst VerdictShortlist

Updated Nov 12, 2026

Deepgram is one of the most established independent voice AI platforms, and its streaming speech recognition, text-to-speech and agent stack is used in production by companies like Twilio, Cresta, Decagon and NICE. Most buyers building real-time transcription or voice agents should include it in their evaluation. Test accuracy on your own audio against AssemblyAI and ElevenLabs, and get roadmap commitments on the Daily/Pipecat integration.

Signal Provenance

Latest observation date and source for each signal category

Signal CategoryLast UpdatedSource
Funding Signals
Last funding round, Funding stage
Nov 12, 2026Cited web research
Hiring Velocity
Headcount, Headcount growth
Nov 15, 2026Research agents (estimate)
Technical Output
Research papers, GitHub activity
Nov 15, 2026Hugging Face Hub + news feed + research agents
News Sentiment
News mentions, News sentiment
Nov 15, 2026Verified news feed
Partnership Signals
Enterprise partnerships, Distribution partners
Nov 12, 2026Research agents (estimate)
Revenue Indicators
ARR estimate, Revenue growth
Nov 15, 2026Research agents (estimate)
Leadership Signals
CEO tenure, Founder-led
Nov 15, 2026Research agents (estimate)
Competitive Position
Category rank, Market share
Nov 15, 2026Research agents (estimate)

Why this Bluescore?

6 dimensions, each scored 0–10 against all 250 companies. Each adds its weight × score to the Bluescore — see methodology

Deepgram's Bluescore of 45.7 rests most on Talent & Research (6.6, 72nd percentile) and Customer Adoption (5.4, 67th percentile). Its weakest area is Business Model & Revenue (3.6, 20th percentile).

In plain English

Deepgram's Bluescore of 45.7 is lifted by Talent & Research, where it scores 6.6 of 10 at the 72nd percentile among the AI250, and Customer Adoption, at 5.4 of 10 and the 67th percentile. It is held back mainly by Business Model & Revenue, which scores 3.6 of 10 at the 20th percentile and leaves 13.5 of its 21 possible points unclaimed, and by Data & Model Moat, which scores 3.5 of 10 at the 22nd percentile with 12.4 of 19 points left. Business Model & Revenue would move the score most, since it carries a 21% weight and has the most points still available.

Written by AI from the scores below and checked against them; updated nightly.

D1 · Business Model & Revenue (21%)
3.6
+7.5 of 21 pts · 20th pct
D2 · Product Depth (21%)
5.0
+10.4 of 21 pts · 54th pct
D3 · Data & Model Moat (19%)
3.5
+6.6 of 19 pts · 22nd pct
D4 · Talent & Research (16%)
6.6
+10.6 of 16 pts · 72nd pct
D5 · Customer Adoption (13%)
5.4
+7.0 of 13 pts · 67th pct
D6 · Investment Momentum (10%)
3.6
+3.6 of 10 pts · 31st pct

Traction Signals

Last Round

Series C

Risks

Updated Nov 12, 2026

Strong competition from ElevenLabs, AssemblyAI, OpenAI realtime models and the major cloud providers
Medium
Integrating the Daily acquisition could slow delivery of the new Voice Agent Platform
Medium
Independent benchmarks show competitors beating Nova-3 on word error rate and price
Medium
Rapid model releases and API changes, such as the dg-token callback removal, need ongoing engineering upkeep
Low
Pipecat is moving to an independent foundation, so future vendor neutrality is unproven
Low
Feedback