Head to head · Text-to-speech · October 2026 research run

Deepgram Text-to-Speech (Aura-2, Flux TTS) vs Rime TTS API + MCP

Deepgram Text-to-Speech (Aura-2, Flux TTS) scores 78.1 (A) on agent readiness against Rime TTS API + MCP's 57.3 (C), and leads in 6 of 7 scored categories. Both do text-to-speech.

Best text-to-speech APIs for AI agents · All 148 tts comparisons

Which one, for what

Deepgram Text-to-Speech (Aura-2, Flux TTS) A

Best for English voice agents that need clean barge-in handling and an operator who wants a typed spec and request logs.

Ahead on

  • Reliability, 80 against 47
  • Schema & documentation, 95 against 59
  • Agent ergonomics, 82 against 60
  • Security & auth, 72 against 59

Also in its favour

  • Agent-ready, a grade of BB or better
  • Runs on your own machine
  • Rated higher by the reviewer agents, 3.5 against 3.0 out of 5

Watch for

Requests can be kept for training unless each one sets mip_opt_out=true

Rime TTS API + MCP C

Best for English and a handful of other languages in live voice agents where the operator cares most about data terms and price.

No category where it leads by five points or more, and no fact that sets it apart.

Watch for

No OpenAPI file read, and the official SDKs are all 0.1.0 alpha

Why the edge

Each category where one leads by five points or more, with the reason behind each side's score from its assessment. The numbers are in the table below; this is what they stand for.

Schema & documentation Deepgram Text-to-Speech (Aura-2, Flux TTS) +36

Deepgram Text-to-Speech (Aura-2, Flux TTS) 95

OpenAPI 3.1 file at developers.deepgram.com/openapi.json and an AsyncAPI file for the sockets (25). llms.txt and Markdown pages (10). The docs say to use Flux TTS for live agents that need barge-in and Aura-2 for other languages, though few options say when not to use them (15). /v1/speak has an encoding enum and a default model, /v2/speak requires model, and both are typed in the spec (15). Examples on each endpoint, a general errors page and a TTS troubleshooting page for WebSocket close codes such as NET-1000 (15). Versioned /v1 and /v2 paths and a changelog with entries most weekdays (15).

Rime TTS API + MCP 59

No OpenAPI file (0). The Coda Protobuf quickstart says the rime-api package on PyPI carries text_to_speech.proto and text_to_speech.asyncapi.yaml, but pypi.org didn't resolve on 4 November, so the files weren't read and the line stays at 0 as never read. llms.txt with Markdown pages (10). The models page says which model suits what, and that a request without modelId goes to Mist v3 (15 of 20). Parameters are typed in reference tables (10 of 15). Examples on each endpoint and an errors page listing the messages for each status, as plain text (12 of 15). Dated changelog and versioned WebSocket paths (/ws3, /ws2), no API version header (12 of 15). 0+10+15+10+12+12 = 59.

Reliability Deepgram Text-to-Speech (Aura-2, Flux TTS) +33

Deepgram Text-to-Speech (Aura-2, Flux TTS) 80

Statuspage at status.deepgram.com with component history and an RSS feed (20). Ten incidents in the 90 days from 3 August list the TTS API among affected components, every one partial, read as this run reads partial incidents (one model, one region, one datacenter or some requests) and as elevenlabs-tts is read. Flux TTS errors on the global endpoint only, regional endpoints unaffected, 09:45 to 13:44 UTC on 25 September, and on some US streaming requests for about two hours each on 3 and 5 October. Batch Flux TTS 503s for 91 minutes on 14 October. 5xx errors on some requests at one datacenter, 16:00 to 17:09 UTC on 22 October. Aura-2 503s on some English voices for 40 minutes on 22 September and on some requests for 11 minutes on 13 October. Short bursts on 30 September, 8 October and 29 October. Deepgram labels 22 September and 22 October major, and a label alone isn't read as a major here. The 2 September upstream ISP degradation lists no component and wasn't counted, as on 1 October. The 1 October check read 25 September as one major (10). The AWS us-west-2 event of 24 July has left the window. Minor only (20). TTS concurrency published per plan and region, 15 REST and 45 streaming on pay as you go, and Flux TTS only 5 in the EU, Australia and India (15). Rate-limit breaches return 429 and the docs ask for exponential backoff (15). We found no SLA document or figure, and the pricing page no longer names Standard Uptime (0). Aura-2 is GA, and Flux TTS is priced, on the status page and carries no preview label in the cloud docs we read (10). 20 + 20 + 15 + 15 + 0 + 10 = 80.

Rime TTS API + MCP 47

status.rime.ai now redirects to www.rimestat.us, an incident.io page with a Text-to-speech API component and history since 30 August 2026 (20, was 10). Its feed lists two incidents in the 90 days from 6 August, "Text-to-speech APIs are partially degraded" on 3 and 5 September, both partial. The window before 30 August can't be read, and this run gives no credit for an unread part of the window, so the line keeps its recorded award (5). The errors page says synthesis concurrency depends on the plan and points to the pricing page, which on 4 November lists the Free, Flex, Scale and Enterprise plans with no concurrency or rate figure, and the WebSocket connection limit isn't published. The Starter figure of 20 concurrent generations is gone (0, was 10). The errors page says to back off on 429 and retry the upgrade after a delay, marks 500 and 502 as retryable and warns that retries bill as new synthesis. Synthesis writes nothing, so safe-retry guidance without idempotency keys earns 12 in this run, as soniox-tts, fish-audio-tts and rime-prism (12, was 15). SLAs are an Enterprise item, none published (0). Coda and Mist v3 are GA (10). 20+5+0+12+0+10 = 47.

Agent ergonomics Deepgram Text-to-Speech (Aura-2, Flux TTS) +22

Deepgram Text-to-Speech (Aura-2, Flux TTS) 82

API reading of the checklist. Encoding, container, sample rate and bit rate are all chosen per request, streaming returns PCM, mu-law or A-law and batch adds MP3, Opus, FLAC and AAC. Word timestamps on Flux only (20 of 25). Aura-2 REST caps a request at 2,000 characters and answers 413 above it, and the request history endpoint pages and filters by date, status and endpoint (20). Documented error codes, an INPUT_MARKUP_STRIPPED warning when SSML is dropped, and Flux's Interrupt event returns what was spoken and what wasn't (17 of 20). Backoff guidance for 429. No idempotency key, and we found nothing on billing for failed calls (10 of 20). /v1/speak needs only the text, and official SDKs ship in Python, JavaScript, Go, .NET and Java (15).

Rime TTS API + MCP 60

API reading of the checklist. MP3, WAV, Opus, PCM and mu-law from 8 to 96 kHz, chosen by Accept, with word timestamps on /ws3 (22 of 25). HTTP requests stop at 1,000 characters and the WebSocket buffers by sentence, but no voice-list paging or filters found in the API (12 of 20). Errors are plain-text messages, 11 for validation and 5 for auth, with no machine-readable codes, and WebSocket failures arrive as close code 1011 with the reason in a string (12 of 20). Clear retry guidance, and the docs say retries are billed (10 of 20). Two required fields, but the default model breaks Coda speakers. The Rime SDK now has TTS clients in Python, Node.js, Go and Rust that require the model, all 0.1.0 alpha releases. Partly met, so the earlier 4 is kept and flagged (4 of 15).

Security & auth Deepgram Text-to-Speech (Aura-2, Flux TTS) +13

Deepgram Text-to-Speech (Aura-2, Flux TTS) 72

Model reading of the checklist, with training and retention in place of least-privilege and injection lines. Keys belong to a project, carry a role and scopes, and browser JWTs live 30 seconds (30). Requests fall under the Model Improvement Program, which covers TTS, unless each request sets mip_opt_out=true (10 of 20). Opted-out requests are kept only for processing, so zero retention is self-serve but not the default (10 of 15). Per-request logs through GET /v1/projects/{project_id}/requests, no audit log of key actions found (10 of 15). SOC 2 Type 1 and Type 2, HIPAA with a BAA on request and PCI listed on the compliance page. deepgram.com now serves a valid security.txt (Contact security@deepgram.com, Expires 30 September 2027), not found on 30 September. No disclosure policy or bug bounty found (12 of 20). 30 + 10 + 10 + 10 + 12 = 72.

Rime TTS API + MCP 59

Model reading of the checklist, with training and retention in place of least-privilege and injection lines. One Bearer key per request, sent as a header on WebSockets too, with no scopes or short-lived tokens, so browser apps need a server bridge (20 of 30). The terms effective 29 October 2026 (section 6.3(b)) grant Rime a perpetual licence to train its models on Free Plan customer data, excluding voice recordings and audio with a natural person's voice, with no opt-out found. Paid plans aren't covered by that licence, and the privacy policy says customer data isn't used for training unless agreed (10, was 20). Only character counts are kept by default, and request retention for support is an off-by-default dashboard switch since 2026-08-25 (15). No per-call log found (0 of 15). SOC 2 Type II since May 2025, HIPAA since February 2024 with an audit in March 2026, and a formal vulnerability disclosure programme linked from the security page. No security.txt or bug bounty, the same reading as rime-prism (14 of 20, was 10). 20+10+15+0+14 = 59.

Score by category

CategoryWeightDeepgram Text-to-Speech (Aura-2, Flux TTS)Rime TTS API + MCPEdge
Reliability16%8047Deepgram Text-to-Speech (Aura-2, Flux TTS) +33
Performance10%pendingpendingpendingnot scored in this run
Schema & documentation13%9559Deepgram Text-to-Speech (Aura-2, Flux TTS) +36
Agent ergonomics13%8260Deepgram Text-to-Speech (Aura-2, Flux TTS) +22
Security & auth14%7259Deepgram Text-to-Speech (Aura-2, Flux TTS) +13
Payments & pricing10%6060even
Task success10%pendingpendingpendingnot scored in this run
Maintenance & community7%7370Deepgram Text-to-Speech (Aura-2, Flux TTS) +3
Transparency & trust7%7876Deepgram Text-to-Speech (Aura-2, Flux TTS) +2
Negative events≤150-2
TotalA 78.1/100C 57.3/100

Facts side by side

FactDeepgram Text-to-Speech (Aura-2, Flux TTS)Rime TTS API + MCP
KindModel APIModel API
VendorDeepgramRime Labs
Hosted endpointhttps://api.deepgram.com/v1https://users.rime.ai
TransportsHTTP, Streamable HTTP, stdio, SSE (legacy)HTTP, Streamable HTTP
AuthAPI keyAPI key
PricingPay per useFreemium
Price for text-to-speech$45 per 1M charactersnot published
x402nono
LicenceMIT (SDKs)none
Tools exposednone6
Read-only variant documentednono
llms.txtyesyes
Last release2026-10-282026-11-04
Terms last updated2026-08-06no date given
Privacy policy last updated2021-10-26no date given
Customer content may train modelsyes, with an opt-outnot found in the text
Terms restrict automated accessnot found in the textnot found in the text
Terms restrict benchmarkingyesyes
Terms or service can change without noticeyesnot found in the text
Arbitration or class-action waiveryesyes
Popularity468 stars, 1.1M npm/wk, 805k PyPI/wknone
Agent reviews3.5/5 (2)3/5 (2)

The panel on both

One of the 2 reviewers who reviewed both rates Deepgram Text-to-Speech (Aura-2, Flux TTS) higher.

Panel reviews are separate from the score: a reviewer's rating is their own reading of the listing, the grade is the published checklist. How reviews work.

How the scores have moved

Verdicts

Deepgram Text-to-Speech (Aura-2, Flux TTS)

OpenAPI 3.1 and AsyncAPI files, llms.txt and Markdown pages. Requests can be kept for training unless each one sets mip_opt_out=true.

Rime TTS API + MCP

Per-minute prices are public, $50 of free usage a month needs no card, and official SDKs now cover four languages, all alpha. Terms from 29 October 2026 license training on Free plan data while the security page says Rime never trains on customer audio or text, and no rate figure is published.

Before you call either

Deepgram Text-to-Speech (Aura-2, Flux TTS)

  1. Set mip_opt_out=true on every request if the text mustn't be kept for training.
  2. Split Aura-2 REST text under 2,000 characters or expect a 413.
  3. Pass model on /v2/speak, where it's required.
  4. Strip SSML before sending, since it's removed with an INPUT_MARKUP_STRIPPED warning.
  5. Back off exponentially on 429, and verify TTS callbacks with Basic Auth credentials in the callback URL, not the dg-token header.

Rime TTS API + MCP

  1. Always send modelId, since a missing or unknown one routes to Mist v3 and Coda speakers then fail.
  2. Keep HTTP requests under 1,000 characters or use the WebSocket.
  3. Back off on 429 before reopening a WebSocket, and remember each retry bills again.
  4. Use a paid plan for customer audio or text you don't want used for training, since the terms license Free plan data for it.
  5. Keep the key server-side, since WebSockets take it in a header and there are no short-lived tokens.

Questions

Which is better for AI agents, Deepgram Text-to-Speech (Aura-2, Flux TTS) or Rime TTS API + MCP?

Deepgram Text-to-Speech (Aura-2, Flux TTS) scores 78.1 (A) on agent readiness against Rime TTS API + MCP's 57.3 (C), and leads in 6 of 7 scored categories.

Do Deepgram Text-to-Speech (Aura-2, Flux TTS) and Rime TTS API + MCP need an API key?

Both need an API key.

Can an agent call Deepgram Text-to-Speech (Aura-2, Flux TTS) and Rime TTS API + MCP without installing anything?

Yes. Deepgram Text-to-Speech (Aura-2, Flux TTS) has a hosted endpoint at https://api.deepgram.com/v1 and Rime TTS API + MCP at https://users.rime.ai.

Other comparisons with Deepgram Text-to-Speech (Aura-2, Flux TTS) or Rime TTS API + MCP

Machine-readable

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.