RXed AI News

Your AI intelligence briefing
SCORE 7.2/1018/20 elements Audited 2026-07-10 · RXed table v1.0

ElevenLabs

The audio benchmark grew into a serious agents platform with a testing framework bigger vendors lack — just watch the shared credit meter, overages bite without warning.

SpecialistVoice & SpeechText-to-SpeechAutomation & AgentsMusicFreemium
Vendor
ElevenLabs · elevenlabs.io
Origin
GB — London
Pricing
Free €0 (10k credits, no commercial license) · Starter ~€6 · Creator ~€20 · Pro ~€92 · Scale ~€277 (3 seats) · Business ~€915 (10 seats) per month; Enterprise custom. One shared credit pool across all products; overages auto-bill on Creator+
Users (official only)
User counts not disclosed; official: $500M+ ARR (Q1 2026), 530 employees (source, 2026-05-05)

Element scores

Reactive
Retrieval & Memory
Orchestration
Validation
Models
Primitives
Pr7
Prompts
Em5.5
Embeddings
Cx7
Context
Tr7.5
Tracing
Lg6.5
LLM
Compositions
Fc8
Function calling
Vx6.5
Vector store
Rg7.5
RAG
Gr6
Guardrails
Mm9
Multimodal
Deployment
Ag8
Agents
Ft7.5
Fine-tuning
Fw7.5
Frameworks & harnesses
Ev8.5
Evaluations
Sm7
Small models
Emerging
Ma7
Multi-agent
Sy
Synthetic data
Pc7.5
Protocols
In
Interpretability
Th5.5
Thinking models
Tap or hover any element to see why it got that score.

Strengths

Best-in-class audio across the board — Eleven v3 TTS out of alpha (02/2026), Scribe v2 STT, Music v2, Dubbing v2 — with G2's independent natural-voice score placing it ahead of Google Cloud TTS. The Agents platform is the surprise: hosted RAG, mature tool calling with MCP support, agent transfers, and an evaluation/testing framework (scenario, tool-call, simulation, probabilistic runs) that most bigger platforms don't match. Momentum is official: $500M Series D at $11B (02/2026), $350M→$500M+ ARR in four months.

Honest dings

Billing is the documented sore spot: Trustpilot sits at ~3.0/5 (~1,005 reviews) driven almost entirely by billing complaints — automatic overages on Creator+ hit without real-time alerts, and one shared credit pool means a minute of music (900 credits) or dubbing (up to 10,000 credits/min) silently eats your TTS budget. The free tier carries no commercial license. No first-party LLM means agent reasoning quality is inherited, and voice cloning remains a misuse surface despite verification gates.

Best for: Teams shipping voice — audiobooks, dubbing, agents for support/sales lines — who want frontier audio quality plus a tested, deployable agent stack, and who will set a calendar reminder to check credit usage before each billing cycle.

Sources

Every audit lists the research it rests on — transparency and traceability are the product. Tools evolve: each audit is a snapshot of its audit date, and re-audits supersede older versions (kept below for reference).