Sort replies: Relevant Recent Liked
Replying to @philipkiely
Does this mean I can use my GLM key in Codex?
6
11
51,072
Not exactly -- open model inference is served by Baseten, and this is available through the new OpenAI Marketplace for enterprise customers.
3
30
36,738
more replies
Replying to @philipkiely
How can non enterprise users do this???
3
2
38
30,311
You can self-serve a similar setup with Baseten Switch: github.com/basetenlabs/baset…

GitHub - basetenlabs/baseten-switch: Local macOS app and gateway for routing AI coding harnesses...

Local macOS app and gateway for routing AI coding harnesses through Baseten - basetenlabs/baseten-switch

github.com
3
7
132
24,205
more replies
Replying to @philipkiely
does the commit cover every model Baseten hosts or just those two?
1
3,268
Any model, just picked those two as examples
1
2,549
Replying to @philipkiely
You rock, Philip!
1
2
9,173
Trying my best, thank you
2
8,003
Replying to @philipkiely
Commenting so I can read this later, but what about those of us that want to use those via consumer accounts?
1
4
13,942
You can self-serve a similar setup with Baseten switch: github.com/basetenlabs/baset…

GitHub - basetenlabs/baseten-switch: Local macOS app and gateway for routing AI coding harnesses...

Local macOS app and gateway for routing AI coding harnesses through Baseten - basetenlabs/baseten-switch

github.com
1
13
11,288
more replies
Replying to @philipkiely
Amazing
1
5
13,318
I'm about to build so much stuff with Codex
9
10,648
Replying to @philipkiely
Wow, never expected frontier labs to add opensource models to their products. What about support in Agents API?
1
90
20,498
Supports Responses API and Codex today
51
15,444
Replying to @philipkiely
@grok what does counts against ai commit means
1
1
1,887
Replying to @philipkiely
OpenAI enterprise customers, how about Pro subscriptions? Would be huuuge to have it
2,562
Replying to @philipkiely
Cool but open models dont work as well in codex. theres a pretty big difference using them in open harnesses usually
63
Replying to @philipkiely
@philipkiely I hope it doesn't have safeguards in place to do thorough pentesting on our own apps ?? It defeats the purpose you know.. 😅
215
Replying to @philipkiely
@grok Does baseten offer EU data residency or inference?
1
663
Replying to @philipkiely
Why cant i just have my local model in codex and keep yours let them work together? Codex can bring them in, but only by removing yours (and yours still drain)
1
1,898
Replying to @philipkiely
I need to be able to freely switch between multiple ChatGPT accounts within Codex; if one account hits its 5-hour limit, I can switch to another. 我需要能在codex中多个chatgpt 账号能自由切换，一个账号的5小时限额满了的时候可以切换另外一个
1
362
Replying to @philipkiely
@grok Explain how to set this up, let's say to use GLM5.3-Flash in Codex Desktop UI
1
802
Replying to @philipkiely
I thought codex has been supporting other providers for a while
466
Replying to @philipkiely
So you're telling me that the models we can run locally should be run on Codex? Are you serious? LOL
1,881
Replying to @philipkiely
now to convince my employer to use this…
2
1,932
Replying to @philipkiely
Wow. This is awesome. Does Basten have abliterated versions of GLM 5.3 and others? If not feature request.
373
Replying to @philipkiely
Will OpenAI or you train on the user sessions?
1,000
Replying to @philipkiely
Also add recently mentioned abliterated GLM5.3 for offensive cyber featured by Anthropic, deleted by huggingface, rereleased by Audn @ huggingface.co/audnai/pencla…

audnai/penclaw-GLM-5.3-abliterated · Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

huggingface.co
2
391
Replying to @philipkiely
nice
4,448
Replying to @philipkiely
Solid
2,623
Replying to @philipkiely
Can teams compare models on the same repo task with the same tools and tests? Cost per accepted change would be useful alongside token price
46
Replying to @philipkiely
Well obviously they have to access to other models, when opus 5.5 is still better than sol 6.1
611
Load more