Union Alpha on OpenRouter: Tests and Reviews

A Polish user notes that the free stealth model Union Alpha showed up on OpenRouter just as Ox Alpha did a month earlier, and wonders which model is behind it this time.

A roundup of Union Alpha's 256K context, multimodal and agentic coding positioning and low-cost rumors draws a parallel to Ox Alpha turning out to be GLM-5.3-Flash and guesses it is an anonymous GLM-5.4 test.

One user claims Union Alpha is just a rough router that mixes Llama 3.3 70B, Qwen 3.6 35B-A3B and GLM 5.3 Flash, falling back to OpenRouter's GLM 5.3 Flash under load.

The same prompt was given to Union Alpha, DeepSeek V4.1 Flash and Muse Spark 1.3 to build an interactive neural network visualization, with the code behind each result compared.

After two hours of failed requests on OpenCode, the user got the pelican-on-a-bicycle SVG prompt to run in 26 minutes and rated the model's quality as solid.

Union Alpha is available with unlimited tokens on Mercury Cloud and can be paired with Claude Code, OpenCode or Mercury Code.

A developer says Union Alpha navigates a heavy C++ project with ease and understands context fast, but OpenRouter errors every couple of seconds make benchmarking impossible.

A user says Union Alpha does not support thinking mode and still takes very long on the most basic tasks, calling the experience unbearable.

A Japanese user who tested both says GLM's safeguards are looser than Union Alpha's and easier to work with.

The blogger says they had not been following benchmarks and just noticed Union Alpha, and plans to test whether it can compete with GPT-6 Astra on UE5.

A tutorial video shows how to connect the free Union Alpha model on OpenRouter to Claude Code and runs a pelican test to show its first output.

The author reports Union Alpha runs at 300-400 tokens per second in Codex with 262k context and image input, free for a week, and compares it to Ox Alpha, which was revealed as GLM-5.3-Flash after a week.