OpenAI just dropped a WILD benchmark. 👀
On Agents’ Last Exam, GPT-6.2 Astra scored 5.5 points higher than Claude Fable 5.2 at just 5% of the cost, according to OpenAI.
One benchmark isn’t the whole story, but the cost difference is insane.
Astra or Fable: which one are you using?
2
3
241
🚨 BREAKING: OpenAI just released GPT-6.2 Astra and GPT-6.2 Sol.
What’s new:
→ Astra sets new highs on professional and research evals
→ At less than 1/3 the cost per task of GPT-6 Astra
→ Better at websites, slides and 3D designs
→ +22 points on cyber benchmarks, so it ships with tighter safeguards
Available now in Codex, the API and ChatGPT Work (Plus, Pro, Enterprise, Edu).
Is this enough to make you leave Claude Code?
Some of you have been asking very politely. Some of you have just been asking.
GPT-6.2 Astra and Sol are now available.
Super efficient and built to perform where it counts—the real world.
openai.com/index/introducing…
1
84
Introducing Jev @typesafeai powered jobs-api
Try it at treg.to/jobs-api
- One API to any jobs data
- $0.03 to run a full job hunt
- Real salary & Hiring manager data
- Plugin to any agent for your dream job
Fully open source, 0% markup
Git Repo below 👇
I read 50 Head of Content job posts.
They all want the same 5 things (and it's not "creativity"):
1. A system, not a viral post
2. Proof you can grow from zero
3. Taste: knowing what NOT to publish
4. Distribution, not just creation
5. Content tied to revenue
Being good at content gets you noticed.
Owning the numbers gets you hired.
1
2
219
GPT-6 Luna just came surprisingly close to Astra in an OpenAI test.
On HealthAdminBench, which measures healthcare administrative tasks:
-GPT-6 Astra: 94.3% of available points
-GPT-6 Luna: 91.8%
The surprising part? OpenAI says Luna cost 98% less, at $0.085 per task.
Both were tested at maximum reasoning effort. This is one benchmark, not proof that Luna is better for every task.
3
6
474






