In September 2026 I announced a series of Gemini updates on LinkedIn, covering the new Gemini 4 Argon model, expanded token limits, pricing, Managed Agents features, TTS voice replication, and partner SDK releases.
Gemini 4 Argon supports up to 1M output tokens, up from 64K.
Live agents are priced at $0.005/min input and $0.018/min output.
Managed agents preview (antigravity‑preview‑09‑2026) delivers up to 30% lower costs with up to 22% higher cache hits.
Gemini 3.8 TTS can replicate a voice from 30s of audio, or describe one in a sentence.
The Artificial Analysis Quality Index score is 82.6 (#1).
These announcements reflect decisions to boost model capacity, improve security with the Credentials API, and lower operational costs through caching.