Joined October 2022
- Tweets 4,617
- Following 1,705
- Followers 14,263
- Likes 43,422
943 Photos and videos
Pinned Tweet
Jun 13
Seems like @Zai_org actually took the criticism to heart.
GLM 5.2 is:
> Notably faster than GLM 5.1
> I tested it with 6 bug fixes and an implementation on OpenCode; everything went through cleanly with solid planning.
> It's also good for DB and system design stuff when you pair it with the right agent skills.
Jun 13
Fable gets banned today. And the same day @Zai_org drops GLM 5.2.
Wild timing.
Honestly, I'm trusting the Chinese providers more than the American ones at this point.
13
6
315
29,580
23h
It can't even find the goddamn architecture/README.md files, bro.
Opencode and Claude handle this in seconds without breaking a sweat.
5
551
Jun 13
Seems like @Zai_org actually took the criticism to heart.
GLM 5.2 is:
> Notably faster than GLM 5.1
> I tested it with 6 bug fixes and an implementation on OpenCode; everything went through cleanly with solid planning.
> It's also good for DB and system design stuff when you pair it with the right agent skills.
Jun 13
Fable gets banned today. And the same day @Zai_org drops GLM 5.2.
Wild timing.
Honestly, I'm trusting the Chinese providers more than the American ones at this point.
13
6
315
29,580
Jun 13
It just completed a full end-to-end field removal in one shot. DB schemas, validation schemas, services, routes, frontend pages, components, unit tests, integration tests, and e2e tests... all of it. At once. Wild.
2
1
13
2,484
Jun 13
Chinese and American AI lab marriages might be what saves us.
Americans build the models, and the Chinese open-source them.
That's the playbook. That's how we win.
Jun 13
I am willing and able to marry any Anthropic MTS who needs their Mythos access back
14
1,150
Jun 13
Fable gets banned today. And the same day @Zai_org drops GLM 5.2.
Wild timing.
Honestly, I'm trusting the Chinese providers more than the American ones at this point.
Intelligence should be open, accessible, and ready to build with, empowering every developer, everywhere.
GLM-5.2 is now available to all GLM Coding Plan users, including Lite, Pro, Max, and Team plans.
docs.z.ai/devpack/latest-mod…
As our new flagship model, GLM-5.2 delivers powerful coding capabilities, usable 1M-context support, and continued strengths in long-horizon tasks.
API and Chatbot services will launch next week. The model will also be officially open-sourced next week under the MIT License.
The future of AI is open, and it belongs to the people.
18
5
152
41,270
Jun 13
1M context window, by the way.
Currently stress testing it on a massive project. Will report back once I see how it handles the load.
1
7
1,289
Alper Tunga retweeted
Jun 9
GLM 5.1 has been surprisingly solid and genuinely fun to work with.
Here's the skill stack I've been running it with:
> Mental models & proper thinking: cc-thinking-skills
Specifically, "thinking-first-principles", "thinking-systems", and "thinking-feedback-loops"
> Debugging: systematic-debugging
> Schema design with PostgreSQL: postgresql-table-design
> Queues with RabbitMQ: rabbitmq-expert
> Caching with Redis: redis-core, redis-connections, redis-query-engine
All the skill links below 👇
Introducing GLM-5.1: The Next Level of Open Source
- Top-Tier Performance: #1 in open source and #3 globally across SWE-Bench Pro, Terminal-Bench, and NL2Repo.
- Built for Long-Horizon Tasks: Runs autonomously for 8 hours, refining strategies through thousands of iterations.
Blog: z.ai/blog/glm-5.1
Weights: huggingface.co/zai-org/GLM-5…
API: docs.z.ai/guides/llm/glm-5.1
Coding Plan: z.ai/subscribe
Coming to chat.z.ai in the next few days.
5
2
27
4,328
Alper Tunga retweeted
Jun 11
some of you asked about the tech stack and infra behind @NodeStoryAI
frontend: react @vite_js for the dashboard and @nextjs for the landing page
backend: express.js
caching and pub/sub: redis via @outplanecom
db: postgresql 17 via @outplanecom
ai: gemini, chatgpt, nano banana 2, seedance 2, kling 3, and many more via @eachlabs
hosting: @outplanecom
object storage: @Cloudflare R2
domain: @spaceship
package manager: @bunjavascript
auth: @better_auth
Jun 11
A year ago, my team and I were tired of the same loop...
Type a prompt, get a result, start completely over. Every campaign, every asset, from scratch.
So we built @NodeStoryAI, a visual canvas for connecting prompts, models, and media into reusable workflows. Build it once. Swap inputs. Run again.
Image gen, video gen, 20 models, all in one place.
Team seats for agencies. API access when you need it.
Solo creators stop rebuilding pipelines and actually start shipping.
We built this as a team of 5 (@TugserOkur @merveeksi6117 @fatih_dev_chef @wkaandemir) because we needed it ourselves and hope it helps you too.
nodestory.ai
2
3
21
2,665
Alper Tunga retweeted
Jun 11
A year ago, my team and I were tired of the same loop...
Type a prompt, get a result, start completely over. Every campaign, every asset, from scratch.
So we built @NodeStoryAI, a visual canvas for connecting prompts, models, and media into reusable workflows. Build it once. Swap inputs. Run again.
Image gen, video gen, 20 models, all in one place.
Team seats for agencies. API access when you need it.
Solo creators stop rebuilding pipelines and actually start shipping.
We built this as a team of 5 (@TugserOkur @merveeksi6117 @fatih_dev_chef @wkaandemir) because we needed it ourselves and hope it helps you too.
nodestory.ai
6
8
34
6,730
Jun 11
A year ago, my team and I were tired of the same loop...
Type a prompt, get a result, start completely over. Every campaign, every asset, from scratch.
So we built @NodeStoryAI, a visual canvas for connecting prompts, models, and media into reusable workflows. Build it once. Swap inputs. Run again.
Image gen, video gen, 20 models, all in one place.
Team seats for agencies. API access when you need it.
Solo creators stop rebuilding pipelines and actually start shipping.
We built this as a team of 5 (@TugserOkur @merveeksi6117 @fatih_dev_chef @wkaandemir) because we needed it ourselves and hope it helps you too.
nodestory.ai
6
8
34
6,730
Jun 11
The tech stack:
x.com/altudev/status/2065214…
Jun 11
some of you asked about the tech stack and infra behind @NodeStoryAI
frontend: react @vite_js for the dashboard and @nextjs for the landing page
backend: express.js
caching and pub/sub: redis via @outplanecom
db: postgresql 17 via @outplanecom
ai: gemini, chatgpt, nano banana 2, seedance 2, kling 3, and many more via @eachlabs
hosting: @outplanecom
object storage: @Cloudflare R2
domain: @spaceship
package manager: @bunjavascript
auth: @better_auth
4
700
Jun 11
some of you asked about the tech stack and infra behind @NodeStoryAI
frontend: react @vite_js for the dashboard and @nextjs for the landing page
backend: express.js
caching and pub/sub: redis via @outplanecom
db: postgresql 17 via @outplanecom
ai: gemini, chatgpt, nano banana 2, seedance 2, kling 3, and many more via @eachlabs
hosting: @outplanecom
object storage: @Cloudflare R2
domain: @spaceship
package manager: @bunjavascript
auth: @better_auth
Jun 11
A year ago, my team and I were tired of the same loop...
Type a prompt, get a result, start completely over. Every campaign, every asset, from scratch.
So we built @NodeStoryAI, a visual canvas for connecting prompts, models, and media into reusable workflows. Build it once. Swap inputs. Run again.
Image gen, video gen, 20 models, all in one place.
Team seats for agencies. API access when you need it.
Solo creators stop rebuilding pipelines and actually start shipping.
We built this as a team of 5 (@TugserOkur @merveeksi6117 @fatih_dev_chef @wkaandemir) because we needed it ourselves and hope it helps you too.
nodestory.ai
2
3
21
2,665
Jun 11
We went with @outplanecom for deployment because @rcperdg made testing, staging, and shipping apps incredibly easy for both B2Bs and teams like ours.
They also have a $5K Runway program for startups. If you're looking for a clean deployment setup, I'd definitely suggest checking them out.
1
6
355
Jun 11
Over the past month, @TugserOkur and I have been running @cursor_ai 3 and MiniMax on @opencode pretty much daily.
Composer 2.5 and MiniMax Highspeed 2.7 have been the combo for quick fixes and shipping new features fast. They've been genuinely helpful.
1
5
279
Jun 11
Dig deeper, and you'll see you can customize panel layouts, art styles, and all the details to build your own workflows from scratch.
1
8
330
Jun 11
And our personal favorite... the Product Image Generator node. You feed in your product, character, environment, and style references, and it generates professional product shots that actually look like they came from a studio. This one gets used constantly.
8
307