I was wrong. I said we were a year away from Fable 5 on our desk.
That day is today
An open model BETTER than Fable 5 in some benchmarks just dropped
Better than ChatGPT 5.6 on FrontierSWE. Better than Fable 5 on Automation Bench
This fundamentally changes the AI race forever
If people can start running Fable 5 on their desk, unlimited and for free, they are not going to pay thousands for subscriptions
Now let's be clear: will the hardware you need to run Kimi K3 be obtainable for most? No it won't
It will require multiple very expensive Nvidia chips or a bunch of Mac Studio 512gbs
But look at the slope, not the Y intercept. Over the last year local AI has become SIGNIFICANTLY more efficient and required way less compute
The smartest brains in the world are all attacking the compute problem as hard as they can
This is another step in that direction. And within a year or two, you'll be able to run this on a Mac Mini
Local AI has arrived, and it's not going anywhere
Introducing Kimi K3: Open Frontier Intelligence
🔹 2.8 Trillion Parameters, 1 Million Context, Native Multimodal
🔹 Kimi Delta Attention enables up to 6.3x faster decoding in million-token contexts
🔹 Attention Residuals deliver ~25% higher training efficiency at <2% additional cost
🔹 Built for long-horizon agentic coding and self-evolving workflows
Kimi K3 is now live on on Kimi.com, Kimi Work, Kimi Code, and the Kimi API.
Open Weights by July 27, 2026.
🔗 API: platform.kimi.ai
🔗 Tech blog: kimi.com/blog/kimi-k3
TyinTech retweeted
The Kimi k3 benchmarks are absolutely insane.
DeepSWE is just behind Fable 5, and the Terminal benchmark is ahead of Opus 4.8. Open source is no longer lagging six months behind Western closed-source models.
Read that again, and think about what it all means. I’ll write a detailed post about it later, but just let that sink in.
We got open source 5.6/Fable 5 model from china. Today everything changed.
Big news: Kimi-K3 by @Kimi_Moonshot is now #1 in the Frontend Code Arena with 1679 pts, surpassing Claude Fable 5.
This is a 17-place jump from Kimi-k2.6 (#18 -> #1).
In Frontend, Kimi-K3 ranked #1 in 6 of 7 domains: Brand & Marketing, Reference-Based Design, Data & Analytics, Consumer Product, Simulations, and Content Creation Tools, landing #2 only in Gaming behind Fable 5.
The full model weights will be released by July 27.
Congrats to the @Kimi_Moonshot team on this major milestone!
The biggest cheatcode right now for using AI is
Buy $20/mo Devin @cognition
Spam the hell out of GLM 5.2 (free with pro tier $20/mo)
GLM 5.2 is insanely good "for the price"... you can have it running legitly all day and night
GLM 5.2 does NOT have ability to view images, so switch to Kimi K2.7 when you need to include images in your prompt... then switch back to GLM 5.2 when ready
Use and abuse it before they take it away from being free
TyinTech retweeted
The most valuable skill sets on the planet right now:
1. people who can set up agents properly, manage them, and run local AI models
2. marketers who know how to build distribution
3. robotics engineers who can do all three: build the hardware, wire in the AI, and source manufacturing etc
4. curators who are good at yapping and can do short form video in their sleep
5. the builder-distributor. The one person who can both ship the product AND get it in front of people
6. IRL community builders
TyinTech retweeted
GLM 5.2 might be the “ChatGPT moment” for local AI
The moment many of us see the value in local models
GLM 5.2 isn’t perfect, but it’s really good
1M token context window so it holds an entire codebase at once. Top open model on coding right now. MIT licensed with zero restrictions.
And it runs on your own machine through Ollama or LM Studio.
When you're using it, it kinda feels like Opus 4.8.
I honestly can't believe how good it is.
2027 is probably the year of local AI.
TyinTech retweeted
All day using GLM 5.2. Didn't miss much. First open model that passes the bar as a daily driver. Things are not going to be the same.
Damn, now I want to buy some serious hardware.
When you have limited credits because you're a brokie, choose the best Claude model (Opus 4.8 or Fable or whatever)
THEN
Go to Kimi K2.7 and GLM-5.2 when you run out of credits fo freeeeee
SUPER GOOD, and did I mention FREE!!
Infinite credit, infinite code
Double check the output from Kimi and GLM with Opus 4.8++ later just to be safe
Anthropic shut down Fable 5 yesterday... While they stole the attention @Kimi_Moonshot secretly released K2.7-Code in the same day
- 1T MoE
- 32B active
- 256K context
MCP Mark Verified: 81.1
Claude Opus 4.8: 76.4
MLS Bench Lite: 35.1 (tied with GPT-5.5)
30% fewer reasoning tokens!!!
KernelBench-Hard: MoE kernel regressed 0.222 > 0.157
Free (good) beats rate-limited (great) for many teams when taking into account unit economics
Personally, when Opus 4.8 runs out (quickly...), I switch to Kimi K2.7 and it does a good jobs picking up the slack... FOR FREE!
Tried Kimi yet?
TyinTech retweeted
🌘Kimi-K2.7-Code Weights & code
We explore endless possibilities together with the community🤝
huggingface.co/moonshotai/Ki…
Product analytics and marketing analytics are not the same thing
Most founders install Google Analytics on day one and never realize they're measuring traffic, not behavior
Here's the difference 👇🏽
GA4 tells you where people came from... It does not tell you why they left as easily
If your onboarding has five steps and users vanish at step three, GA4 shows you a "bounce rate."
That number is useless for fixing the leak
@posthog tracks business events, not page views. The events that matter are the ones that map to revenue or retention:
→ signup_completed
→ onboarding_step_reached
→ extension_installed
→ voice_generation_completed
→ checkout_initiated
These are decisions, not impressions.
The funnel I actually care about:
Landing → Signup → Onboarding Step 3 → Extension Install → First Generated Output → Paid Conversion
A funnel tool shows me exactly where people drop. A traffic tool shows me... colors on a map
Session replays caught a bug in which no error log flagged... A user on Firefox couldn't install the Chrome extension. Silent failure. They clicked, nothing happened so they left
I watched the replay in Posthog. CSP issue in my middleware. Fixed it
Without replay, I would never have known
Here's how I split my events... Steal the taxonomy:
Server (via PostHog Node SDK):
- signup, onboarding_progress, extension_install, voice_generation, checkout, churn_signal
Client (via PostHog JS):
- feature_discovery, time_to_first_value, help_article_viewed
Extension (via telemetry API):
- platform_detected, injection_success, generation_triggered
The free tier is actually free. Not a trial. Not capped at ten events. I run production analytics for multiple apps on the free plan
The paid plan costs fractions of a penny per event. The day I outgrow the free tier is the day the business justifies it
If your primary analytics tool is built for marketers, you are flying blind as a product builder
Switch to an event-based funnel tool & build an instrumentation layer that answers product questions and not traffic questions
Your retention curve will thank you 🙏🏽
MoSt PoWeRfUl YeT!?!?
Anthropic just released their "safeguarded version of Claude Mythos" in "Claude Fable 5" today...
At a whopping 2x the price of their highest OPUS 4.8 model
These insane god-mode level AI's are costing an arm and a leg... thus making it even more of a "rich get richer" type play
This is way out of price range for the "normies" and really only within the top 5%'s hands at this point
I keep saying opensource is catching up, but the gap is still so huge from the "best" tiers now like Fable 5 and the open source models...
Will play a bit with it, maybe 1hr before all my weekly credits on my $200/mo tier run out...
Whats your take on the new model so far?
TyinTech retweeted
Chrome extension just learned how you write.
Now every platform gets your voice. X. LinkedIn. Email. Reviews. Everywhere.
One Brand DNA. Infinite on-brand posts.
That's the Zebrafy difference. 🦓
TyinTech retweeted
AI content? All sounds the same. Bland. Generic. Robot soup.
Why? No soul. No Brand DNA.
Zebrafy scrapes your site + socials. Extracts your voice. Chrome extension drops Generate + Score everywhere.
Stop blending in.
getzebrafy.com 🦓
I created an auto job seeker app in a matter of hours 👇
-Scrapes X, LI, Indeed and other job boards for specific terms/job titles using @apify
- Finds hiring managers/emails if they have public
- Creates (on brand using @GetZebrafy ) a job-specific resume, DM, and email (auto sends if email)
- Saves into database for analytics, updates, etc
- Automated daily self-email with stats
- Original link for the job posting
How would you improve/make this better? 🤖
22 days left and counting 🚨
Check your brand score and get told how your brand is. Free on the site. All you have to do is enter your username for both X and LinkedIn
TyinTech retweeted
I just joined @GetZebrafy's Founders List — AI that writes in YOUR brand voice, not robot voice.
Launching May 1st. Join with my link and we both get rewards 🦓 getzebrafy.com/prelaunch?ref…


























