It's been a very busy 9 days for Qwen! Qwen3-235B-A22B-Instruct-2507 – 21st July
Qwen3-Coder-480B-A35B-Instruct – 22nd July
Qwen3-235B-A22B-Thinking-2507 – 25th July
Qwen3-30B-A3B-Instruct-2507 – 29th July
Qwen3-30B-A3B-Thinking-2507 – today
INNOVATION
-
Qwen3 Releases Five Major Model Updates in Nine Days
By
–
-

Qwen3-30B Models Generate Pelican Bicycle Art Comparison
By
–
On the left the pelican on a bicycle from Qwen3-30B-A3B-Thinking-2507, on the right the pelican from Qwen3-30B-A3B-Instruct-2507
-

Qwen 3 30B Model Successfully Generates Functional Space Invaders Game
By
–
Not a great pelican, but it did write me a functional version of space invaders (unlike yesterday's non-thinking 30B-A3B which produced a game that didn't quite work) https://
simonwillison.net/2025/Jul/30/qw
en3-30b-a3b-thinking-2507/
… -
Self-improving AI systems and automatic task generation
By
–
It's not quite *self*-improving, as it still requires the user to describe the task, but of course, yeah, in theory, the system could also auto-generate tasks.
-
Single-turn to multi-turn generation expansion announced
By
–
Right now, single-turn tasks only, but I will very likely expand this to multi-turn generation soon!
-
Codestral 25.08 and Complete Mistral Coding Stack Launch
By
–
Meet the new Codestral 25.08 and the Complete Mistral Coding Stack for Enterprises!
-
Training AI Models for Under One Dollar Now Possible
By
–
Yep! And you can train a model for literally $1 or less
-
$22.5B Valuation Achievement Credited to Jeb Leadership
By
–
Not sure we'd have a $22.5B valuation to celebrate if it weren't for Jeb.
-
Vibe Training Replaces Vibe Coding in AI Development
By
–
Forget vibe coding. Vibe *training* is now a thing.
-
OpenRouter Credits Enable Cost-Effective AI Model Execution
By
–
For this run, I spent $0 on the GPU, and 40 cents on OpenRouter credits for prompt generation, RULER ranking, etc.