Short answer: the AssemblyAI Startup Program grants up to 200,000 hours of free speech-to-text and voice AI API usage, one of the largest credit pools in the voice AI space, to companies at Series B or earlier that have raised less than $10M and are building voice AI with a launched or imminent product. There is no separate application form on the public site; you apply by contacting AssemblyAI directly through its startup program page.
What the program gives you
| Element | Detail |
|---|---|
| Credit | Up to 200,000 hours of free API usage |
| Funding stage | Series B or earlier |
| Total raised | Under $10M |
| Product stage | Launched, or imminent launch |
| Beyond credit | Co-marketing and community perks |
| Approval difficulty (our tracking) | Low |
We track the current terms on the AssemblyAI perk page in the Perkstack catalog, last verified in May 2026. That catalog listing itself already ranks around position 6.4 in search with real impressions, which is the direct evidence that founders are actively looking for this program and not finding a dedicated explainer.
Why AssemblyAI's credit pool is unusually large
Speech-to-text and voice AI usage is billed by the hour or minute of audio processed rather than by token, which makes a 200,000-hour pool a genuinely large amount of runway compared to a dollar-denominated grant from a text-based LLM provider. For a startup building anything that listens (transcription, realtime streaming, voice agents, conversation intelligence), that volume can cover a meaningful stretch of both development and early production traffic before a bill starts. The catch, and the reason the eligibility bar exists, is that AssemblyAI is sizing this for companies that are specifically building voice AI as a core product, not using transcription as an incidental feature.
Eligibility, read carefully
Three criteria define the bar:
- Series B or earlier. Later-stage companies fall outside the published range.
- Under $10M raised total. This is a lifetime funding cap, not a per-round figure, so a company that has raised several smaller rounds adding up past $10M is outside the window even at an early stage.
- Building voice AI with a launched or imminent product. AssemblyAI's own framing centers on companies where speech and voice are the core of what they are building, not a side feature bolted onto an unrelated product.
The $10M lifetime cap is the detail most likely to catch out a company that otherwise looks early-stage. A voice-first startup that closed an unusually large seed round can exceed the funding threshold even while still pre-Series-A in structure.
How to apply
AssemblyAI's startup program does not run through a public self-serve form the way some cloud programs do. The process:
- Confirm your funding stage and total raised against the Series B, under $10M criteria before reaching out.
- Have your product status ready to describe: launched, or a specific, credible near-term launch date. AssemblyAI is evaluating whether voice AI is genuinely the core of what you are shipping, not a peripheral feature.
- Go to AssemblyAI's contact page for the startup program and submit your company and product details directly.
- Describe your use case concretely. Since this is a direct contact process rather than a form-based application, a clear, specific description of what you are building with AssemblyAI's API (transcription, streaming, voice agents, conversation intelligence) is what a reviewer on the other end has to work with.
- Wait for a response. AssemblyAI does not publish a fixed review timeline for this program on its public page.
What the credit actually covers
The 200,000-hour figure applies to AssemblyAI's speech-to-text and voice AI API usage: transcription, realtime streaming, and the conversation intelligence features built on top of the core transcription engine. For context on where AssemblyAI's models sit against competing speech-to-text providers on raw price, see the cheapest AssemblyAI Universal pricing table, which tracks AssemblyAI against Deepgram and other hosts in our weekly verified rankings. A large free-hours pool from the startup program is worth more or less depending on how AssemblyAI's underlying rate compares to alternatives once the credit runs out, which is exactly what that comparison table is for.
Comparing voice AI options if you do not qualify
If your company falls outside the Series B or $10M threshold, or your product does not center on voice AI specifically, a few alternative paths exist for cheap or free speech-to-text access:
- Compare current per-minute pricing directly. Cheapest Whisper large v3 API and cheapest AssemblyAI Universal API both track live, weekly-verified rates across hosts, including DeepInfra's especially low Whisper pricing.
- Check general-purpose inference credit programs that are not voice-specific but still offset early API spend broadly, covered in the startup credits checklist.
Why a large free-hours pool matters more for voice than for text
Speech-to-text billing is structurally different from LLM text billing in a way that makes a large hours-denominated credit unusually durable. A text model's cost scales with token count, which varies enormously by task, prompt length, and output verbosity, so a dollar-denominated credit can burn down at very different rates depending on what you build. Audio transcription cost scales far more predictably with the actual duration of audio processed, which is a number most voice products can estimate accurately from expected call volume or recording length well before launch. That predictability means a founder evaluating the AssemblyAI program can forecast, with reasonable confidence, roughly how many months of expected usage 200,000 hours actually represents, in a way that is harder to do with a dollar figure against an LLM workload where usage patterns are still being discovered.
For a startup at the application stage, this is worth doing before you apply rather than after: estimate your expected monthly audio-processing volume based on your product's usage pattern (call center minutes, meeting recordings, voice agent turns), and check that number against the 200,000-hour pool to understand roughly how much runway the credit represents for your specific product, rather than treating the headline number as an abstract "large amount of free stuff."
Bottom line
The AssemblyAI Startup Program is one of the largest credit pools available to an early-stage voice AI company: up to 200,000 hours of free API usage for companies at Series B or earlier with under $10M raised and a launched or imminent voice AI product. There is no public self-serve application; you apply by contacting AssemblyAI directly with your funding and product details. Current terms are tracked on the AssemblyAI perk page in the Perkstack catalog. Create a free Perkstack account to track this alongside 200 plus other verified startup perks.
Related reading: cheapest AssemblyAI Universal API, cheapest Whisper large v3 API, the startup credits checklist.
rest of this guide
The rest of this guide picks up at "What the credit actually covers".
- What the credit actually covers
- Comparing voice AI options if you do not qualify
- Why a large free-hours pool matters more for voice than for text