jack

jack

my quick posts!

so GitHub leaked GPT-5 well before its being announced by OpenAI (still not yet), and it’s interesting there are multiple different versions of it

Although it makes it less confusing imo 😂

Excited to play with these models later!

so GitHub leaked GPT-5 well before its being announced by OpenAI (still not yet), and it’s interesting there are multiple different versions of it

Although it makes it less confusing imo 😂

Excited to play with these models later!

The image is an announcement email from GitHub dated August 6, 2025, declaring that GPT-5 is now generally available in GitHub Models. The release is labeled as a 1-minute read. The top section features a comparison of AI models: OpenAI GPT-5, Llama 4 Scout, and Cohere v2, highlighting their performance metrics on a task to add categories to a transaction prompt. GPT-5 shows a higher accuracy with 90%. Beneath the models is a detailed text explaining GPT-5’s advanced capabilities, including improvements in reasoning, code quality, and user experience. It supports complex coding tasks with minimal prompting. Various GPT-5 model variants are introduced: gpt-5 for logic tasks, gpt-5-mini for cost-sensitive applications, gpt-5-nano for low latency, and gpt-5-chat for enterprise conversations. Users can access these models via the Models tab, GitHub Models playground, or GitHub API, encouraging experimentation with sample prompts. Links are provided for these options.

Spending 5m this morning (whilst family getting ready) remotely accessing my old PC at home and installing the OpenWebUI Windows app hoping that I can access it over Tailscale.

I had my other instance of OpenWebUI up for ages, but when setting up my PDS this was cleared.. :(

Spending 5m this morning (whilst family getting ready) remotely accessing my old PC at home and installing the OpenWebUI Windows app hoping that I can access it over Tailscale.

I had my other instance of OpenWebUI up for ages, but when setting up my PDS this was cleared.. :(