Resources // Voice & Multimodal AI
Vapi for Startups
- VC-only
Offer 90,000 free minutes
Suits for VC-backed
Updated Jun 2026
The offer
- Status: "Applications on hold" as of mid-2026 (the page lists benefits but is not accepting new applications right now).
- 90,000 free Vapi platform minutes, 7,500 minutes/month for 12 months (vs 100 min/mo for regular users). The card's old "$2,000 in credits" is wrong; the offer is in minutes.
- Dedicated private Slack with Vapi forward-deployed engineers; priority and early access to new features
Who qualifies
- Startups building voice-AI products; pre-seed to Series A with $250k+ in funding; launching to customers within ~6 months
- Apply via the form (when open). "Meeting the criteria does not guarantee acceptance." (Older pages said "founded within 5 years / $100k-$5M", that is stale.)
Community Insights
Vapi is the developer-first, mix-and-match voice-agent platform. You wire your own STT, LLM, and TTS, add tools, warm transfers, and CRM actions, and it is praised for flexibility, docs and an active community. The trade-offs: it is more complex than its rivals (Retell is easier; Bland is stronger on multi-prompt flows), real-world latency and goal-completion can disappoint early (one builder cited ~20% goal completion, worse on multilingual calls), and all-in per-minute cost stacks well above the headline once STT, TTS, LLM, and telephony are added. The startup offer is 90,000 free minutes (the offer is in minutes, not $2,000 of credits), and applications are currently on hold.
Best Practices (from community tips)
- Pick Vapi if you have dev resources and want rich automations (CRM, warm transfers, custom tools). Choose Retell for fast no-code, Bland for complex branching.
- Prototype on a few real calls first. Test latency, recognition and goal-completion before scaling. Specify language up front for multilingual.
- Model all-in cost (~$0.20-0.33/min realistic once STT, TTS, LLM, and telephony stack), not the $0.05 platform fee.
- Design graceful human fallback and warm transfers. Expect 20-30% improvement after prompt-tuning cycles.