ElevenLabs published v2.79.0 of its Python SDK on 2026-11-16, adding a tool builder that defines agent tools as code. The same release updates the README to feature the Eleven v4 models, with Eleven v4 described as the company’s most expressive, highest quality speech synthesis model and Eleven v4 Turbo as its most expressive real-time model (ElevenLabs). For developers building voice agents, the release shows the SDK moving agent configuration into code, alongside a workflow builder that arrived in an earlier release.
What v2.79.0 contains
According to the release notes, v2.79.0 has four changes: reporting of unhandled client tool calls to the application, the agent tools as code feature, README documentation for the Eleven v4 models, and a regeneration of the SDK. The notes credit the tool builder to a named contributor and mark the unhandled-tool-call change as the first contribution from an ElevenLabs triaging bot account. A separate release entry dated 2026-11-16 covers the Fern-generated SDK regeneration on its own.
Reporting unhandled client tool calls matters for agent code: an application can now be told when a tool request arrived that nothing handled.
The models the README now features
The README lists Eleven v4, identified as eleven_v4, as the company’s most expressive and highest quality speech synthesis model, and says it supports 90+ languages. Eleven v4 Turbo, identified as eleven_v4_turbo, is described by the README as the company’s most expressive real-time model, with a median latency of ~100ms. Eleven Flash v2.5, identified as eleven_flash_v2_5, supports 32 languages, according to the README.
These are the company’s own descriptions and latency figure, and the README quote does not tie them to an independent measurement. Read side by side, the entries show Eleven v4 and Eleven v4 Turbo both listed with 90+ languages, with Eleven v4 Turbo aimed at real-time use and Eleven Flash v2.5 listed at the narrower count of 32 languages.
Release history over the past weeks
| Date | Entry, as written by the project | Source page | Read on |
|---|---|---|---|
| 2026-11-16 | release v2.79.0: What’s Changed / Report unhandled client tool calls to the application by @elevenlabs-triaging[bot] in #931 / feat: Add agent tools as code (tool builder) by @PaulAsjes in #930 / docs: feature Eleven v4 models in the README by @PaulAsjes in #933 / SDK regeneration by @fern-api[bot] in #935 / New Contributors / @elevenlabs-triaging[bot] made their first contribution in #931 / Full Changelog: v2.78.0…v2.79.0 | https://github.com/elevenlabs/elevenlabs-python/releases | 2026-11-19 |
| 2026-11-16 | release fern-generation-base–fernapi–fern-python-sdk: [fern-generated] SDK regeneration | https://github.com/elevenlabs/elevenlabs-python/releases | 2026-11-19 |
| 2026-11-12 | release v2.78.0: What’s Changed / fix: preserve Gemini answer text in speech engine streams by @vinsymoyya in #929 / SDK regeneration by @fern-api[bot] in #932 / New Contributors / @vinsymoyya made their first contribution in #929 / Full Changelog: v2.77.0…v2.78.0 | https://github.com/elevenlabs/elevenlabs-python/releases | 2026-11-19 |
| 2026-11-06 | release v2.77.0: What’s Changed / [Speech Engine] Add auth parameter to create and update by @SamuelGD in #921 / feat: Add agent workflows as code (workflow builder) by @PaulAsjes in #922 / SDK regeneration by @fern-api[bot] in #924 / New Contributors / @SamuelGD made their first contribution in #921 / Full Changelog: v2.76.0…v2.77.0 | https://github.com/elevenlabs/elevenlabs-python/releases | 2026-11-19 |
| 2026-11-04 | release v2.76.0: What’s Changed / Add entity_redaction to the realtime speech-to-text wrapper by @JacekDuszenko in #912 / SDK regeneration by @fern-api[bot] in #918 / Full Changelog: v2.75.0…v2.76.0 | https://github.com/elevenlabs/elevenlabs-python/releases | 2026-11-19 |
| 2026-10-27 | release v2.75.0: What’s Changed / Report conversation disconnection details by @gauravch-code in #907 / SDK regeneration by @fern-api[bot] in #909 / New Contributors / @gauravch-code made their first contribution in #907 / Full Changelog: v2.74.1…v2.75.0 | https://github.com/elevenlabs/elevenlabs-python/releases | 2026-11-19 |
| 2026-10-20 | release v2.74.1: What’s Changed / Forward enable_logging in the custom music compose_detailed wrapper by @lpastusz in #894 / Bump version to 2.74.1 by @lpastusz in #902 / Full Changelog: v2.74.0…v2.74.1 | https://github.com/elevenlabs/elevenlabs-python/releases | 2026-11-19 |
| 2026-10-20 | release v2.74.0: What’s Changed / SDK regeneration by @fern-api[bot] in #901 / Full Changelog: v2.73.1…v2.74.0 | https://github.com/elevenlabs/elevenlabs-python/releases | 2026-11-19 |
The project’s release list shows frequent releases, with agent and speech features arriving in most of them. The v2.78.0 release on 2026-11-12 preserved Gemini answer text in speech engine streams. The v2.77.0 release on 2026-11-06 added an auth parameter to Speech Engine create and update, plus agent workflows as code. The v2.76.0 release on 2026-11-04 added entity_redaction to the realtime speech-to-text wrapper. The v2.75.0 release on 2026-10-27 reported conversation disconnection details. On 2026-10-20 the project shipped v2.74.1, which forwards enable_logging in the custom music compose_detailed wrapper, and v2.74.0, a regeneration.
The table shows that tool and workflow builders landed in v2.77.0 and v2.79.0, so agent definition in code is a thread across more than one release. The music wrapper fix in v2.74.1 is the only music-specific entry in these rows; the others cover speech, agents, transcription, or plain SDK regeneration (ElevenLabs).
Licence and contribution rules
The repository’s licence is MIT, as reported by the GitHub API. The README says the library is generated programmatically, so changes made directly to it would have to be moved into the company’s generation code. Contributors who want to change SDK behaviour should expect their work to be reworked upstream rather than merged as submitted.
What the sources we read do not state
We check six facts on every story in the full text of each source. The pages listed under Sources do not state hardware requirements.
How we built this
Release historyDated release entries published by the project
Dated entries read by our code from the project's own release notes, news list or changelog, newest first, with the page each was read from; the entry text is the project's own wording, not a check of what it says. Each entry is shown whole. An entry of a release feed is the release's title and its notes, whole lines joined by " / ". Rows: the newest 8 of 10.
Sources
The sources this article was written from. Links open the original page.


