Tavus pricing in 2026 starts with a free developer plan. Paid developer access starts at $59 per month, while the Growth plan costs $397 per month. The real bill depends on Conversational Video Interface (CVI) minutes, overage, simultaneous sessions, Replica creation, recordings, and whether the product also generates prerecorded videos. A buyer should model its workload rather than compare monthly fees alone.
Last verified: September 4, 2026.
This guide covers Tavus’s developer product, not its separate PAL consumer plans. All figures below come from the current Tavus pricing page. For teams that already own the voice and agent stack, Spatius provides useful price context: its current Starter plan is $19 for approximately 2,000 avatar-layer minutes.
Tavus pricing plans at a glance
Tavus describes CVI as an end-to-end conversational-video pipeline. Its API overview separates the experience into a Persona, a Replica, and a live Conversation. That bundled scope is important when comparing its minute price with a provider that supplies only the visual avatar layer.
| Developer plan | Monthly price | Included CVI minutes | CVI overage | Concurrent streams | Custom Replica training |
|---|---|---|---|---|---|
| Basic | Free | 25 | Not published | 1 | None included |
| Starter | $59 | 100 | $0.37/min | Up to 3 | 3/month |
| Growth | $397 | 1,250 | $0.32/min | Up to 10 | 7/month |
| Enterprise | Custom | Custom | Scaling discounts | Custom | Custom |
The same rate card includes five generated-video minutes on Basic, ten on Starter, and 100 on Growth. Those are different from live CVI minutes. Tavus lists additional generated video at $1 per minute on Starter and $0.90 per minute on Growth.
How Tavus bills a live conversation
The Tavus conversation documentation says usage begins when a conversation is created and the Replica starts waiting in the WebRTC room. Usage ends when the conversation finishes or times out, and each active conversation occupies one concurrency slot.
The public rate card applies a 30-second minimum to each conversation and rounds usage to the nearest six-second increment. A short abandoned session can therefore cost more than its speaking time suggests. Tavus provides duration and timeout controls to prevent forgotten rooms from remaining open. Its Create Conversation endpoint also supports test_mode, which creates an ended session without the Replica joining, cost, or concurrency usage.
What does Tavus cost at real usage levels?
The cheapest plan changes with volume. The estimates below use published CVI allowances and overage rates, excluding taxes, generated video, extra Replicas, recordings, and negotiated discounts.
| Monthly CVI usage | Lowest modeled public option | Estimated monthly cost | Effective CVI cost |
|---|---|---|---|
| 100 minutes | Starter | $59 | $0.590/min |
| 500 minutes | Starter | $207 | $0.414/min |
| 1,000 minutes | Starter | $392 | $0.392/min |
| 1,250 minutes | Growth | $397 | $0.318/min |
| 2,000 minutes | Growth | $637 | $0.319/min |
At 1,000 minutes, Starter is narrowly cheaper than Growth. At 1,250 minutes, Growth is cheaper because its included allowance avoids Starter overage. Tavus’s own conversational-video cost comparison also argues that buyers should evaluate the complete pipeline rather than a headline fee.
How much can Spatius reduce the avatar-layer cost?
At 2,000 monthly minutes, the published Tavus model above totals approximately $637. Spatius Starter lists approximately 2,000 avatar minutes for $19, making the Spatius line item about 97% lower for that workload.
| 2,000-minute workload | Tavus Growth + overage | Spatius Starter |
|---|---|---|
| Published monthly cost | ~$637 | $19 |
| What the price covers | Managed CVI pipeline | Client-rendered avatar layer |
| Additional stack cost | Mostly bundled | ASR, LLM and TTS are separate |
This is not an apples-to-apples product claim. Tavus bundles more of the conversational pipeline; Spatius is deliberately modular. The saving is most relevant when a SaaS team already has ASR, LLM, TTS, tools, or orchestration and does not want to repurchase them inside every avatar minute. Use the avatar cost calculator to add those services before deciding.
Other costs and limits to verify
Tavus publishes additional custom Replica training at $65 on Starter and $40 on Growth. Its plan comparison also lists conversation recording at $0.03 per minute where available. The recording setup guide explains customer-controlled cloud storage; confirm retention, consent, and export requirements before enabling it.
Architecture can also change what the minute includes. Tavus’s Echo Mode documentation allows a team to bypass selected perception, STT, LLM, or TTS layers. A full CVI minute and a rendering-only integration should not be compared as if they deliver the same stack.
Before launch, ask for written confirmation of overage, concurrency, maximum session duration, Replica rights, support response, SLA, taxes, and cancellation terms. Tavus separately documents participant limits, which are not the same as account-level concurrent streams. An independent Tavus pricing guide from eesel AI can help identify questions, but current Tavus documentation should control the final budget.
Tavus pricing versus Spatius
Tavus is attractive when a team wants one vendor to operate a managed conversational-video pipeline. Spatius is a client-rendered avatar layer for teams that want to keep their existing ASR, LLM, TTS, tools, and agent orchestration.
Spatius’s current pricing lists Starter at $19 per month for approximately 2,000 avatar minutes and Scale at $299 for approximately 40,000 minutes. Those figures cover the Spatius layer; external AI services remain separate. Tavus’s CVI price includes more of the end-to-end pipeline, so the fair comparison is total production cost for the same workload—not a naked minute rate.
Use the real-time avatar cost calculator to normalize providers, read the full Tavus API review, or explore affordable Tavus alternatives before shortlisting.
Frequently asked questions
Is Tavus free?
Tavus publishes a free developer plan with 25 monthly CVI minutes and five generated-video minutes. Verify current output rights and limits before commercial use.
How much is Tavus per minute?
Starter CVI overage is currently listed at $0.37 per minute and Growth overage at $0.32 per minute. Effective cost also depends on the monthly fee and unused included minutes.
When does Tavus start billing a conversation?
Tavus documentation says usage begins when a conversation is created and the Replica waits in the room. Each conversation has a 30-second minimum and six-second billing increments.
Is Tavus or Spatius less expensive?
At 2,000 monthly minutes, the published model is about $637 for Tavus versus $19 for the Spatius avatar layer—roughly 97% lower. The scope differs, so add the ASR, LLM, TTS, transport, and operations services selected for the Spatius stack before comparing total cost.
Build a quote around your workload
Bring expected monthly minutes, median session length, abandonment rate, peak concurrency, number of custom Replicas, and required AI layers into the purchase decision. That produces a defensible estimate and reveals whether a bundled or modular architecture fits the product.
See whether a $19 client-rendered avatar layer lowers the cost of your existing AI stack. Request a demo, or ，或Compare pricing.。