Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: OSS Agent I built topped the TerminalBench on Gemini-3-flash-preview

github.com

Post reply on HN
Print view View original

161–167 of 167 posts

Re: Show HN: OSS Agent I built topped the TerminalBench on Gemini-3-flash-preview

#161

Post by GodelNumbering »

GodelNumbering wrote: post #127
gchamonlive wrote: post #98

hey there! thanks for the project! I was intrigued with the claims so I wanted to test it myself. First I (vibe)made an AUR package I could use to install it from git source, from master: https://aur.archlinux.org/packages/dirac-cli-git Then I went in to see what's what, but I there isn't support for gemini-cli login, and importing from opencode doesn't work, failing with a message "Something went wrong. Could not re…

Hey thanks for doing this! I will be looking into the gpt versions. It doesn't support gemini CLI because google seems to ban users for using it, there was a big controversy about it some time ago so I decided to leave it alone for now. Also, feel free to reach out to me if you want to discuss anything specific

UPDATE: chatgpt 5.5 using both codex and API are fully supported now

Re: Show HN: OSS Agent I built topped the TerminalBench on Gemini-3-flash-preview

#162

Post by GodelNumbering »

redrove wrote: post #52
GodelNumbering wrote: post #27

Earlier quoted context omitted.

Assuming you logged in with OAuth, I am guessing you are trying to use gpt-5.5? In my tests, it worked using gpt-5.4 for me and I assumed gpt-5.5 is not available to me because I am on the free plan Do you have the subscription that allows 5.5? If so, I can look into what changed in API. Sorry I rarely use openAI so it is a bit of an untrodden path

Yes I'm on ChatGPT Pro (OAuth) and I'm trying to use gpt-5.5-xhigh. That was the issue, 5.4 works just fine. Support for service: priority (GPT /fast mode) would also be cool!

Gpt 5.5 is now fully supported using codex and api

Re: Show HN: OSS Agent I built topped the TerminalBench on Gemini-3-flash-preview

#163

Post by gchamonlive »

GodelNumbering wrote: post #161
GodelNumbering wrote: post #127

Earlier quoted context omitted.

Hey thanks for doing this! I will be looking into the gpt versions. It doesn't support gemini CLI because google seems to ban users for using it, there was a big controversy about it some time ago so I decided to leave it alone for now. Also, feel free to reach out to me if you want to discuss anything specific

UPDATE: chatgpt 5.5 using both codex and API are fully supported now

Neat! I'll give it a try. It'd be nice to try to mix gpt 5.5 with local qwen3.6 to see if the context and retrieval optimizations can alleviate the context limitations of running such model on a consumer card (I have a single 3090)

Re: Show HN: OSS Agent I built topped the TerminalBench on Gemini-3-flash-preview

#164

Post by overfeed »

giancarlostoro wrote: post #157
overfeed wrote: post #155

Earlier quoted context omitted.

Building ML model training and serving infrastructure is real-world engineering. Nevermind the user-facing apps and supporting services. > Sure they can build ML models, but I see how they improve upon them after years, and its always some really old "lesson learned" elsewhere in the industry. There's a thousand projects that make things like Claude Code use less tokens, and edit more efficiently, and nobody at Anthr…

> Building ML model training and serving infrastructure is real-world engineering. Nevermind the user-facing apps and supporting services. Looking at Anthropics status info for the last 90 days only serves to prove that they aren't hiring the right people for the right roles. > They have fully internalized the bitter lesson; the result is they get better returns improving the next model over squeezing out performance…

> Looking at Anthropics status info for the last 90 days only serves to prove that they aren't hiring the right people for the right roles

Look at any[1] dashboard over the past 6 months. It's less about the people working there an more about what leadership is demanding, industry-wide: Productivity* is the only metric that matters now - measured by how quickly teams under them can squirt out new features. Leadership desperately need a win because of the amounts invested, so stability becomes what it is now. Nothing to do with the rank and file, though it's Engineering that will be blamed when the time comes, I hope the multitudes of CTOs are earning enough to justify them being sacrificed to appease shareholders.

1. If your company has a SEV or SLA dashboard, look at it and compare the levels before and after mandated AI-productivity pushes by management.

Re: Show HN: OSS Agent I built topped the TerminalBench on Gemini-3-flash-preview

#165

Post by redrove »

GodelNumbering wrote: post #162
redrove wrote: post #52

Earlier quoted context omitted.

Yes I'm on ChatGPT Pro (OAuth) and I'm trying to use gpt-5.5-xhigh. That was the issue, 5.4 works just fine. Support for service: priority (GPT /fast mode) would also be cool!

Gpt 5.5 is now fully supported using codex and api

cool!

Re: Show HN: OSS Agent I built topped the TerminalBench on Gemini-3-flash-preview

#167

Post by mrwolf »

Thanks for publishing this. I tried it out and it works pretty well, nice job! I had the "feeling" the agent was more effective and accurate, not only cheaper.

The only thing that I had issues with is the retry mechanism. For example, when I get error 429 as the endpoint is being rate limited (using openrouter) the session gets stuck and stops.

This is the most annoying, and most crippling issue I've encountered with all open source agents. Fixing this would be a huge accomplishment

Post reply on HN