Deepgram Startup Program
AI Platform Credits Verified May 2026
Up to $100K in speech AI API credits — STT, TTS, voice agents, diarization (pre-Series A, direct apply)
Up to $100K in speech AI credits — transcription, voice agents, TTS, and speaker diarization for pre-Series A startups
Who qualifies
Every condition below is taken from the vendor’s own published criteria. Read them before you spend an afternoon on the application.
New customers only
Existing or previously paying accounts are excluded. If you are already on a paid plan, ask before you migrate — some vendors make an exception, most do not.
Startups
The programme is aimed at startups. Vendors read this loosely, but expect to describe the company and what you are building on the application.
Up to $100,000 in Deepgram API credits valid for 12 months. Available to pre-Series A startups with under $10M funding, new Deepgram customers, with a live website and company email. Direct application — no VC or accelerator required. Credits cover all Deepgram products: Nova-3 STT, Aura TTS, and voice agent APIs.
About Deepgram Startup Program
Quick answer
The Deepgram Startup Program gives eligible pre-Series A startups up to approximately $100,000 in Deepgram API credits, valid for around 12 months, that can be spent across Nova speech-to-text, Aura text-to-speech, speaker diarization, and the Voice Agent API. You apply directly, credits are loaded into your Deepgram account, and you pay standard metered rates once the allocation is exhausted.
What is the Deepgram Startup Program?
The Deepgram Startup Program is a credit-based offering from Deepgram, one of the most established independent voice AI infrastructure providers, designed to give early-stage companies meaningful runway on its speech-to-text, text-to-speech, and voice agent APIs. Public program materials reference up to roughly $100,000 in API credits distributed over a typical 12-month window, positioning it as one of the more generous voice-AI-specific startup offers available today.
Why a voice-AI-specific credit program matters
Most founders know the big three hyperscaler programs — AWS Activate, Google for Startups Cloud Program, Microsoft for Startups — and they often assume those will cover their voice workloads. In practice, those programs give you generic compute and LLM credits, but the bill for transcribing millions of minutes of audio or running real-time voice agents can dwarf the value of generic cloud credits. A purpose-built program like Deepgram's puts the credit dollars exactly where voice-heavy startups spend them: minutes of audio processed, characters synthesized, and concurrent voice-agent sessions sustained.
Deepgram Startup Program vs alternative voice AI credits
Founders often compare Deepgram's program against generic hyperscaler credits and against rival speech AI vendors. The table below is a high-level, typical-scenario comparison — exact terms shift over time and should always be re-verified at signup.
| Program | What it covers | Typical credit value | Best fit |
|---|---|---|---|
| Deepgram Startup Program | STT (Nova-2/3), TTS (Aura), diarization, Voice Agent API | Up to ~$100K | Voice-heavy startups building agents, transcription, or call analytics |
| AWS Activate | General AWS compute, storage, and many third-party marketplace offers | Varies; up to $100K+ for top tiers | Startups with broad cloud workloads beyond voice |
| OpenAI / Azure OpenAI credits | LLMs, embeddings, image, and (in some cases) Whisper STT | Varies by partner | LLM-first products that only need basic transcription |
| AssemblyAI / Speechmatics startup offers | STT only | Smaller credit pools, often 6–12 months | Pure-transcription products that do not need TTS or voice agents |
Where the Deepgram Startup Program shines — and where it does not
✓ Apply if you:
- Are pre-Series A with under ~$10M raised and a live product or pilot.
- Are building a voice agent, IVR, or conversational AI product on top of Deepgram.
- Need both STT and TTS and want to spend one credit pool across both.
- Process enough audio that paying list rate would meaningfully hurt your margins.
- Value low-latency streaming inference for real-time experiences.
✗ Skip if you:
- Have already raised a Series A or significantly more than $10M.
- Are an existing Deepgram paying customer (the program targets net-new accounts).
- Only need a few hundred minutes per month and can comfortably pay list rate.
- Need a perpetual discount — credits expire, they do not roll over.
Realistic cost scenarios for early-stage teams
How far an up-to-$100K allocation actually goes depends on your product. A real-time voice agent that transcribes and synthesizes audio for every customer call will burn through credits much faster than a meeting-notes app that processes one meeting per user per week. Founders who treat the credit pool as a runway, not a windfall, get the most value: they instrument usage early, pick the right Nova model tier for accuracy vs cost, and route low-value traffic to cheaper endpoints while reserving premium models for revenue-bearing flows.
It is also worth modeling the post-credit world on day one. Once credits expire, Deepgram's standard metered pricing applies, and that is the number that should drive your gross margin assumptions. A startup whose unit economics only work inside the credit window is not actually a viable business yet — and the program is most useful when it accelerates the path to pricing that works at full list rate.
What the credit covers
- Up to $100K API Credits — Credits apply across all Deepgram products including Nova-3 (best-accuracy STT), Aura TTS, voice agent APIs, speaker diarization, and sentiment analysis
- Nova-3 — Best-in-Class STT Accuracy — Deepgram Nova-3 delivers industry-leading word error rates across accents, domains, and noise conditions — outperforming OpenAI Whisper and Google Speech-to-Text in third-party benchmarks
- Real-Time Streaming — Sub-300ms transcription latency for live voice applications — essential for real-time voice agents, live captions, and interactive voice response systems
- Speaker Diarization — Automatic speaker identification — label who said what in multi-speaker recordings. Essential for meeting transcription, interview analysis, and call centre applications
- Text-to-Speech (Aura) — Deepgram Aura TTS produces natural-sounding speech with ultra-low latency — designed specifically for real-time voice agent applications where latency is critical
Programme tracks
Verified May 2026. What Deepgram Startup Program publishes for each stage — confirm on the application, since credit programmes are re-cut more often than list pricing.
| Track | Value | Who it is for |
|---|---|---|
| Startup Credits | Up to $100K | Pre-Series A, under $5M raised, building voice/speech applications |
How to apply
4 steps. The last one is the part most people skip.
- 1
Open Deepgram Startup Program through the link on this page
It carries our referral tag. The terms you get are identical either way, and it never changes what this page says about the programme.
- 2
Have the eligibility evidence ready
Applications are checked against 2 conditions — new customers only, startups. Incorporation date, cap table and a one-line description of what you are building cover most of it.
- 3
Size the migration against the 12 months window
Credits start burning from activation, not from when you get round to using them. Work out what you will genuinely consume in that window before you move production workloads across.
- 4
Know the rate you land on when it runs out
Up to $100,000 in Deepgram API credits valid for 12 months. Available to pre-Series A startups with under $10M funding, new Deepgram customers, with a live website and company email. Direct application — no VC or accelerator required. Credits cover all Deepgram products: Nova-3 STT, Aura TTS, and voice agent APIs.
Where this programme wins and loses
What works
- Best-in-class transcription accuracy
- Nova-3 model beats OpenAI Whisper on accuracy and speed
- includes TTS and voice agent APIs
- excellent for voice-first products
What doesn't
- Requires approval and pre-Series A status
- less useful for non-voice products
- competitive transcription market growing fast
The bottom line
Deepgram is the best speech AI API for production voice applications and $100K in credits from a direct-apply program makes this one of the most valuable AI credit programs available to pre-Series A founders.
Deepgram Startup Program FAQ
The questions we actually get asked about this programme.
Ask us something elsePre-Series A startups with under $10M in funding, a live website, and company email. New Deepgram customers only. No VC or accelerator affiliation required — apply directly at offers.deepgram.com/startup-program.
At Nova-3 pricing ($0.0059/minute), $100K covers approximately 16.9 million transcription minutes. For a typical 60-minute meeting, that is 281,000 meetings. Credits cover all Deepgram products including STT, TTS, diarization, and voice agent APIs.
Deepgram Nova-3 outperforms OpenAI Whisper on word error rate (WER) across most audio domains including medical, legal, and noisy environments. Google Speech-to-Text lags significantly on accented speech and domain-specific vocabulary. Deepgram also delivers much lower latency for real-time streaming than both alternatives.
Yes — and you should. Deepgram handles speech-to-text (user voice in) and ElevenLabs handles text-to-speech (AI voice out). Together they provide the complete voice pipeline for a voice agent at zero cost during development.
Deepgram Nova-3 is a proprietary streaming model optimised for low-latency real-time transcription. OpenAI Whisper is a batch model that excels on clean, pre-recorded audio but has high latency for live applications. For real-time voice agents, Nova-3 wins on latency. For offline transcription of clean audio, Whisper is competitive.
