Tellama v1.3.3
Tellama v1.3.3 makes Qwen3 answer fast.
Highlights
- Qwen3 responds quickly now. Qwen3 base models (like Qwen3 4B) are hybrid — they support both a thinking mode and a fast non-thinking mode. Tellama previously treated them as thinking-only and forced every reply through hidden chain-of-thought, so the first visible word was slow to appear. They now run in Fast mode by default (no hidden reasoning), so answers start right away. Switch to Deep Thinking in Settings when you want full reasoning.
- Genuinely reasoning-only models (QwQ, DeepSeek-R1, Qwen3 Thinking variants) still use Deep Thinking as before.
APK integrity
- File:
tellama-v1-3-3.apk - SHA-256:
6a926e0a9752526fa4fb9cca85039e88ad644b39028813d4a07bd934819d73d4 - Package:
com.tellama.app - Version:
1.3.3(versionCode 33) - Signer certificate SHA-256:
5c3180b0290135491a0dec1dba6acab9119a8e17a60c8663d390bb2073007516
Updating
Tellama v1.2.1 and later check the public release feed and validate the APK SHA-256, package, higher version, and signing certificate before opening Android's installer.