LemonSlice AI pricing starts with a free preview and paid self-serve plans from $8 per month. The real production cost depends on included avatar minutes, overage rates, concurrency, call duration, annual billing, and whether you bring your own voice agent or use LemonSlice’s hosted speech and agent services. Compare those meters before comparing headline subscription prices.
Last verified: October 19, 2026. Prices, credits, model access, and limits can change. Confirm the live LemonSlice pricing page before purchasing.
LemonSlice plans at a glance
The public selector displayed these monthly prices and lower annual-billing equivalents at verification:
| Plan | Monthly billing | Annual-billing equivalent | Public positioning |
|---|---|---|---|
| Free | $0 | $0 | Chat with featured avatars; no custom avatar API use |
| Starter | $8/month | $7/month | Entry API plan |
| Creator | $40/month | $33/month | More included usage |
| Professional | $100/month | $83/month | Higher-volume self-serve use |
| Scale | $240/month | $200/month | Larger production workloads |
| Ultra | $960/month | $800/month | Additional models, longer calls, and higher concurrency |
| Enterprise | Custom | Custom | Contracted scale, governance, and deployment requirements |
The selector can change which allowance details are visible. Do not infer every tier’s minutes or concurrency from the Starter values.
What the Starter price includes
At verification, the Starter selection showed 41 included minutes per month, $0.22 per extra minute, three concurrent calls, and a 30-minute conversation limit. Calls above the concurrency limit were described with a 2x burst multiplier.
Those details matter more than the $8 subscription alone. A workload can exceed included minutes, concurrency, or call duration independently. Model all three.
LemonSlice’s Avatar API page describes the product as a real-time face layer for an existing LLM and voice stack. It listed plans from $8 per month and approximately $0.16 per included base-model minute. The broader pricing page advertised rates as low as $0.048 per minute at scale. These are different contexts, not one universal rate.
The existing LemonSlice AI review explains the LiveKit, Pipecat, and hosted routes. The LemonSlice alternatives comparison covers the architecture tradeoff, while the real-time avatar pricing guide provides a broader normalization framework.
Avatar-only and hosted-agent costs are different
LemonSlice can receive speech from your existing agent and return a synchronized avatar stream. In that route, your team still pays separately for speech recognition, the LLM, text-to-speech, transport, observability, and engineering.
The hosted experience can include VAD, STT, LLM, and TTS around the avatar. Treat that as a broader service boundary. Confirm the current hosted usage rate, included models, tool access, data handling, and whether external providers add separate charges.
| Cost layer | Avatar-only route | Hosted route |
|---|---|---|
| Speech recognition | Your provider | Confirm what LemonSlice includes |
| LLM and tools | Your stack | Confirm models, tools, and limits |
| Text-to-speech | Your provider | Confirm included voice service |
| Avatar rendering | LemonSlice minute or credit model | LemonSlice plus hosted-agent meter |
| Real-time transport | Your LiveKit or Pipecat path | Confirm the bundled transport boundary |
Build a production cost model
Use a representative month rather than the best demo session:
- Estimate completed sessions and average connected duration.
- Include abandoned calls, reconnects, retries, and idle time.
- Model peak simultaneous sessions, not only monthly minutes.
- Add ASR, LLM, TTS, transport, storage, and monitoring.
- Compare monthly and annual commitments using the same workload.
For example, 1,000 five-minute sessions imply 5,000 connected avatar minutes before retries. The correct bill cannot be calculated from the entry subscription alone; it requires the live tier allowance, overage rate, concurrency requirement, and hosted or bring-your-own architecture.
Which LemonSlice plan should you choose?
Use Starter for a limited API proof of concept. Move to a higher self-serve plan when measured minutes or concurrency justify it. Ultra is relevant only when its additional models, longer calls, or higher concurrent-session allowance solve a real requirement. Enterprise fits teams that need contractual scale, support, zero-data-retention options, data residency, or other governed controls.
Do not upgrade from a spreadsheet estimate alone. Run a pilot using the actual source image, voice, devices, session pattern, and interruption behavior. If LiveKit is already part of the stack, review the Spatius LiveKit integration to compare the transport and rendering boundaries.
FAQ
How much does LemonSlice AI cost?
On October 19, 2026, monthly public prices were $8 Starter, $40 Creator, $100 Professional, $240 Scale, and $960 Ultra, with Free and custom Enterprise options. Annual billing showed lower monthly equivalents.
Does LemonSlice offer a free plan?
Yes, but the public page describes Free as chatting with featured avatars. Custom avatar creation and API access require a paid plan.
Is LemonSlice priced per minute?
Plans combine a subscription with included usage and overage pricing. The effective minute cost depends on the selected tier, billing period, and workload.
Does LemonSlice pricing include an LLM and voice?
Not in every route. The Avatar API can sit on top of your existing LLM and voice stack, while hosted experiences can include more speech and agent services.
What hidden costs should teams model?
Model external ASR, LLM, TTS, transport, retries, idle sessions, burst concurrency, monitoring, and engineering. Confirm exactly what the selected plan includes.
Compare the complete stack, not one minute
If you are comparing a cloud-video avatar with a client-rendered visual layer, request a Spatius demo. Bring your expected sessions, concurrency, devices, and current voice-agent architecture so the cost comparison uses the same workload.