DEV Community: RainyTech

StarBot Spark: What Developers Should Check Before Paying a Deposit on a $7,999 Humanoid Robot

RainyTech — Mon, 09 Nov 2026 13:50:19 +0000

Short version: StarBot Spark is a 3-foot humanoid robot from $7,999, or $5,999 for the first 100 confirmed orders. That's far below the $13,500 The Gadgeteer lists for a Unitree G1. But if you want to develop on it, the things you most need to know aren't published yet: the onboard computer, the cameras and sensors, the software kit, and a ship date.

This isn't a hands-on test. I haven't touched a Spark, and every number below comes from StarBot, its press release or The Gadgeteer.

▶ Full video: https://youtu.be/I1dWw6LR-LI

What StarBot says Spark is

StarBot Robotics is based in Santa Barbara, California. It announced Spark on November 4, 2026.

Height: about 3 ft (90 cm)
Weight: about 35 lb (16 kg)
Joints: 22 degrees of freedom
Battery: 24 V, up to 3 hours of runtime, about 40 minutes to charge
Software: over-the-air updates, described as an "open development platform"
Made in: the United States, according to StarBot

StarBot calls it North America's first US-made developer humanoid in its size and price class. That's the company's own claim.

The price, compared

Robot	Starting price	Notes
StarBot Spark (first 100)	$5,999	Introductory price
StarBot Spark	$7,999	3 ft, 22 DoF
StarBot Astra	$9,999	5 ft, 23 DoF
Unitree G1	$13,500	As listed by The Gadgeteer

The $200 reservation deposit counts toward the price. Tax, shipping, delivery and final configuration are confirmed later. Schools, universities and research programs can ask for custom pricing.

What a developer still can't find out

A developer robot is only as good as what you can program on it. As of this writing, these are the open questions:

Onboard computer: not specified.
Cameras and sensors: not specified.
The SDK: it isn't documented in the sources I read. You can't yet tell which languages, APIs or level of joint access you'd get.
Simulation: StarBot's site lists NVIDIA Isaac simulation but doesn't say how it works on Spark.
Warranty and safety certifications: no word yet.
Ship date: none. StarBot's terms say launch timing isn't guaranteed.

There's also a documentation warning sign. The Gadgeteer found that the detailed spec sheet on StarBot's site describes Nova, the company's restaurant robot, and not Spark. The dates are inconsistent too: one page lists October 18 and another says Spark was released in September.

Where StarBot came from

StarBot's earlier robot, Nova, takes restaurant orders by voice, takes card payments and sings. StarBot says it has served more than 20,000 people, at $4,999 plus a $499 monthly subscription.

The company bought a Unitree G1 as a research platform in early 2025 and later built its own humanoids for buyers the G1 prices out. So it has shipped a service robot, but that isn't the same as shipping a programmable platform with a supported SDK.

What I'd ask before putting down $200

If you're considering a reservation for a lab or a classroom, I'd get answers to these in writing:

Which SDK and languages are supported, and is low-level joint control exposed?
What compute and sensors ship in the final configuration?
Can I run my code in simulation before the robot arrives?
What's the refund policy if delivery slips?
What warranty, repair and safety certification apply?

Verdict

The Gadgeteer's take is fair: the G1 is the safer bet with a known schedule, and Spark is the cheaper option for buyers who can wait. If StarBot delivers, a sub-$8,000 humanoid starts to fit a school budget and not just a research lab. Until the SDK and compute details are public, though, I'd treat "developer platform" as a promise and not a spec.

What's the first thing you'd program on a $7,999 humanoid?

Sources

StarBot press release (GlobeNewswire, Nov 4, 2026): https://www.globenewswire.com/news-release/2026/11/04/3395568/0/en/made-in-america-and-priced-at-7-999-starbot-launches-spark-humanoid-robot.html
The Gadgeteer: https://the-gadgeteer.com/2026/11/07/starbot-spark-is-a-three-foot-humanoid-robot-built-in-the-us-for-7999/
StarBot: https://www.starbotusa.com
Full video: https://youtu.be/I1dWw6LR-LI

AI-assisted disclosure: I used an AI assistant to help structure this write-up. This is a research summary of published sources as of November 2026, not a hands-on review. Not sponsored by StarBot or Unitree.

WordCraft: I Tested the Free, Rust-Built Microsoft Word Alternative (203 Pages to PDF in 0.16 s)

RainyTech — Mon, 09 Nov 2026 13:49:39 +0000

Short version: WordCraft turned a 203-page document into a PDF in 0.16 seconds, and Claude Code added a full magazine page to a running document in 79 seconds. It's free, while Microsoft 365 Personal costs $99.99 a year. It also swaps fonts it doesn't have and loses a space when you type over a word, and it's only version 0.16.

▶ Full video: https://youtu.be/t4zeuGFR1Ao

What is WordCraft?

WordCraft is Microsoft Word rebuilt from scratch in Rust by storytold, the team behind ArtCraft.

License: MIT or Apache-2.0
Runs on: Mac, Windows, Linux and in the browser
Files: opens and saves Word files, and exports PDF, Markdown, LaTeX and more
Parity: by its own count, 506 of 521 Word features work (97%). That's self-reported, so treat it as a claim.
Stars: about 17,500 on GitHub, one month after release

I tested version 0.16.0 on a Mac mini M4 with 16 GB of RAM. The download matched its published checksum and was notarized by Apple.

Test 1: Build a real magazine

I skipped the boring letter and made a two-page travel magazine called WANDER:

A full-bleed cover photo using Wrap Text → Behind Text and a 30 cm height
A WordArt masthead in Didot at 110 points
A drop cap on the first paragraph of the story
A sea turtle photo with text wrapped around it, plus a pull quote
File → Export → PDF, done in about a second

All of it worked. One catch: Word's default font, Aptos, isn't on my Mac, so WordCraft warns you and swaps it. The drop cap also overlapped the next letter slightly.

Test 2: Track Changes

Track Changes marks every edit, comments work, and Accept All does what it says. Then I found a real bug:

"48 hours"  ->  "72hours"
"sunrise with"  ->  "dawnwith"

When you double-click a word and type over it, the space after the word disappears. Word puts that space back for you. WordCraft doesn't yet.

Test 3: 203 pages to PDF

This is where Rust earns its keep. I converted a document of 203 pages and 89,532 words from the command line:

0.16 seconds.

Test 4: Let an AI drive it

WordCraft has a built-in MCP bridge, so every button is a command an agent can press. I connected Claude Code to the running app and gave it one job: add a packing list page. Then I took my hands off the mouse.

Result: a headline in the same Didot font, an intro, a packing table and a pull quote
Time: 79 seconds
Tool calls: 26

It also checked its own work before saying it was done. For developers, this is the most interesting part. The document editor becomes something you can automate through commands, while you watch the real UI update.

Verdict

	WordCraft	Microsoft 365 Personal
Price	$0	$99.99/year
203 pages to PDF	0.16 s (CLI)	not tested
Agent control	Built-in MCP bridge	not tested
Maturity	v0.16, small bugs	Decades old

For school work, letters and even a magazine like this one, I'd use it today. If your job lives inside complex Word files, test your own files first. I only built new documents here, so this isn't a verdict on compatibility with complex existing Word files.

Would you let an agent edit your documents, or is that a step too far?

Links

WordCraft on GitHub: https://github.com/storytold/wordcraft
WordCraft website: https://getartcraft.com/apps/wordcraft
Full video: https://youtu.be/t4zeuGFR1Ao

AI-assisted disclosure: I used an AI assistant to help structure this write-up. The tests and timings are my own (Mac mini M4, 16 GB, WordCraft 0.16.0, Claude Code 2.1.317, November 8, 2026). Price from microsoft.com on the same date. Not sponsored by storytold or Microsoft.

GridCraft vs Excel: I Checked 1,968 Formulas From Microsoft's Own Templates in This Free Rust App

RainyTech — Mon, 09 Nov 2026 13:48:44 +0000

Short version: GridCraft matched Excel's saved value on 1,842 of 1,968 formulas across 15 of Microsoft's own templates, and 13 of the 15 files were perfect. It's free, while Microsoft 365 Personal costs $99.99 a year. It also has a real #SPILL! bug that a save can write into your file, and it can't display Khmer yet.

▶ Full video: https://youtu.be/18kJ7HuIvnQ

What is GridCraft?

GridCraft is Excel rebuilt from scratch in Rust by storytold, the team behind ArtCraft and WordCraft.

License: MIT or Apache-2.0
Runs on: Mac, Windows, Linux and in the browser
Features: reads and writes Excel files, with over 500 functions, PivotTables and charts
Stars: about 7,300 on GitHub, one month after launch

I tested version 0.17.0 on a Mac mini M4 with 16 GB of RAM. The download matched its published checksum and was notarized by Apple.

Test 1: 15 real Excel files

Last month's reviews said it opened about half of people's files, so I retested on 15 templates straight from Microsoft: budgets, a loan schedule, a Gantt chart and PivotTables.

The CLI has a useful check for this:

gridcraft-cli check your-file.xlsx

It recalculates every formula and compares the result with the value Excel saved in the file.

One fairness detail: the templates use TODAY(), so my first run showed fake date mismatches. I froze those cells at Excel's saved value so old dates wouldn't count as errors.

Overall: 1,842 of 1,968 formulas match (93.6%)
Perfect files: 13 of 15, including the loan schedule (875/875) and Gantt chart (187/187)

The two failures are very different:

Stocks sheet (expected): it needs Microsoft's online data service, so you get #NAME?.
Budget wheel (a real bug): an old-style table reference, Expenses[Amount], gets implicit intersection in Excel. GridCraft returns #SPILL!, and the chart shrinks to one slice.

Some files also warn about skipped features, such as a PivotTable with calculated items.

Test 2: Save it, then open it elsewhere

There's no Excel on my Mac, so LibreOffice opened every file GridCraft saved. It recalculated 4,117 cells, and only 46 differed, most of them tiny rounding digits.

The bad part: after GridCraft saves the budget wheel, LibreOffice shows #SPILL! too. The save rewrote the formula. Keep a copy of your files before saving over them.

Test 3: 100,000 rows vs LibreOffice

Task	GridCraft	LibreOffice
CSV to Excel file	0.40 s	1.35 s
Recalculate 100,000 formulas	1.83 s	1.06 s

So it isn't faster at everything. The app also opened all 800,000 cells in about 1.3 seconds, and every total matched a separate check in Python.

Test 4: Power features

On a made-up coffee shop's 2026 sales, these all worked: tables, data bars, XLOOKUP, LAMBDA, PivotTables, charts and Solver.

It's still missing three Excel habits: tables don't grow when you add rows, formulas don't fill down as calculated columns, and pivot charts include the Grand Total.

Test 5: Let an AI build the sheet

GridCraft exposes its commands to AI agents. I gave Claude Code one prompt: build a monthly budget for a student in Phnom Penh.

Time: 57 seconds
Tool calls: 31
Result: real formulas, red cells for overspending and a chart

It also checked its own work with screenshots before it stopped.

The test that matters to me: Khmer

I'm in Cambodia, so I made a mini mart sheet in dollars and riel. The exchange-rate formulas worked, but Khmer text and the riel sign (៛) showed as empty boxes, and in the PDF the riel sign turned into a Chinese character. The Khmer text is saved correctly in the file. GridCraft just can't draw it yet.

If your language needs complex text shaping, test that first.

Verdict

For budgets, school work and simple sheets, I'd try it today, and I'd keep a copy of my files. If you need Khmer, or your business lives in Excel, wait a few versions. New releases come every day or two, so these results will age quickly.

What's the one spreadsheet you'd test first?

Links

GridCraft on GitHub: https://github.com/storytold/gridcraft
GridCraft website: https://getartcraft.com/apps/gridcraft
Full video: https://youtu.be/18kJ7HuIvnQ

AI-assisted disclosure: I used an AI assistant to help structure this write-up. The tests and timings are my own (Mac mini M4, 16 GB, GridCraft 0.17.0, LibreOffice 26.8.2, Claude Code with Opus 5.5, November 9, 2026). Price from microsoft.com on the same date. Not sponsored by storytold or Microsoft.

I Pointed Craft Anything at Hacker News. Claude Code Rebuilt It 91% Without Me Writing a Line

RainyTech — Mon, 09 Nov 2026 13:48:35 +0000

Short version: I typed three words, "recreate this website", and Craft Anything rebuilt Hacker News to 91% of the original in 29 minutes and 17.6 million tokens. A copy of 2048 scored only 59% but actually played. Then Deep mode answered questions about the wrong app. It's free and open source, and it's also a very early alpha.

▶ Full video: https://youtu.be/t-GX0THhZfo

What is Craft Anything?

It's a desktop app from storytold, the team behind ArtCraft. You point it at a website, app or game, and say what you want. It then runs a loop you mostly just watch:

Study: opens the original in a private browser and photographs every page at phone, tablet and desktop sizes.
Build: Claude Code builds your version on your own Mac, with your own account.
Check: compares the copy with the original and fixes what it finds.
Ready: you get a project you can run.

It's free, open source (Apache 2.0), written in Rust, and has about 320 tools under the hood. The Mac download is 127 MB, signed and notarized by Apple.

The important default is clean-room mode. It never reads the original's code, only what a normal visitor can see. It copies the design, not the content or the brand.

Test 1: Hacker News

I typed the address, wrote "recreate this website" and pressed start.

First version: 72% like the original
After 16 self-checks: 91%
Time: 29 minutes 34 seconds
Tokens: 17.6 million (about $7.49 at API prices; my Claude plan covered it)

Next to the real site, the layout, orange bar and spacing all match. Look closer, though, and it's called Ticker News, the logo is a T, and every story is made up. That's clean-room mode doing its job. It even listed the things it wouldn't decide for me, such as whether it may use the real name and logo.

Test 2: 2048 from the command line

This time I skipped the GUI: one command, one sentence. Halfway through, my Claude plan hit its five-hour limit and the build stopped. The next morning, one resume command picked up exactly where it left off.

Result: a game called "Fuse": same 4×4 board, new name, colors and font
Score: 59%, because clean-room mode won't copy the art
Time: 17 minutes, 2.8 million tokens

The score undersells it. The tiles slide, the numbers merge, and I made a 64. A low similarity number didn't mean a broken copy. It partly meant the tool refused to copy the art.

Test 3: Deep mode (what broke)

Deep mode takes a program apart and explains how it works. It even offered to download Ghidra, which I declined. I pointed it at PhotoCraft, an open-source app I'm allowed to inspect.

It called the app a zero-byte folder.
I asked how the app is put together, and it answered about Hacker News.
It reported 368 findings. After a restart it showed 1. It had been mixing in my first project.

Expect bugs like this. The app labels itself early alpha, and that's accurate.

What it's good for

Learning how a site is built
Prototyping your own version of an idea
Bringing back old software you own that nobody supports

It also has guardrails: it asks before running commands, and it refuses to bypass logins, DRM or anti-cheat.

Verdict

If you need a real product today, no. It's slow, it burns tokens, and some features break. As a learning tool, though, it's remarkable. Watching an agent measure colors and spacing, score itself, and fix the gaps teaches you a lot about what a page is actually made of.

Plan for the token bill. One website took 17.6 million tokens and one build ran into my five-hour limit, so this isn't something to leave running casually on a small plan.

What would you clone first to learn how it works?

Links

Craft Anything on GitHub: https://github.com/storytold/craft-anything
Full video: https://youtu.be/t-GX0THhZfo

AI-assisted disclosure: I used an AI assistant to help structure this write-up. The tests, timings and token counts are my own (Mac mini M4, Craft Anything 0.3.0, November 2026). Not sponsored by or affiliated with storytold, Hacker News or Anthropic.

I Tested Concat, the Free Open-Source CapCut Alternative (Rust, Offline AI, Built-in MCP Server)

RainyTech — Sat, 07 Nov 2026 00:47:53 +0000

Short version: Concat shrank a 76 MB video to 3 MB in 3.9 seconds, and while it ran I logged every network request: 0 bytes of my video left my Mac. It's free, open source, built in Rust, and it ships its own MCP server so AI agents like Claude Code can edit video. But it's a 10-week-old beta (v0.2.9) with 200+ open issues.

▶ Full video: https://youtu.be/KWJw3RQHElU

What is Concat?

Concat is a desktop video editor that wants to replace CapCut. Everything runs on your own machine.

Free and open source (AGPL-3.0), no account, no watermark, and the site says there is no paid tier.
Built in Rust: about 159,000 lines, first release in September 2026.
Runs on Mac, Windows, Linux, Android and iPhone.
Fast growth: 14,000+ GitHub stars and about 440,000 downloads in ten weeks.
Needs 4 GB of RAM (16 GB recommended for 4K).

The layout will feel familiar if you've used CapCut: library on the left, preview in the middle, inspector on the right, timeline at the bottom. Every panel can be swapped, split or dragged.

Price check (November 2026)

	Price
CapCut Pro	US$19.99/month or $179.99/year
Concat	$0

Test 1: Compress a video (the one I ran myself)

Concat has free tools that run right in the browser. I tried Compress a video in Chrome on my Mac:

Input: 76 MB stock clip, 20 seconds of 1080p H.264
Settings: Balanced quality, 720p, MP4
Result: 3.0 MB (96% smaller) in 3.9 seconds

It encodes on your own GPU, with no server in between. I recorded every network request during the run: 0 bytes of the video were uploaded. The honest flaw: the page still loads Google ads.

Test 2: Auto captions

Captions are one button (the T on the timeline). You pick a Whisper model; the base English one is a 148 MB download, and after that it transcribes fully offline. You get one title per phrase on the timeline, you can restyle all of them at once, and it imports and exports SRT. Their own guide has a section called "why a word is wrong", so expect to fix a few words.

Test 3: Text to speech, voice cloning, background removal

Type a script, pick a voice, and the read lands on your timeline.
Voice cloning from a few seconds of audio (small model 98 MB, studio model about 1 GB).
Background removal for people or objects, plus a smart brush. The person model is 15 MB. No green screen.
Downside: on phones, the built-in voice is still coming.

Test 4: 4K export and the MCP server

Export H.264, HEVC or AV1 up to 4K 60 fps, 10-bit, no watermark at any resolution. This week's update made 4K exports about 2.4x faster.

The part developers will care about: Concat ships its own MCP server. An AI agent like Claude Code gets tools for clips, titles, keyframes, effects and export, the same commands the editor itself uses. Setup is one line in the terminal; the steps are in the repo's docs/MCP.md.

So "cut the silences, add captions, export 1080p" can be a prompt instead of an afternoon.

The catch

Beta: v0.2.9 with 200+ open issues on GitHub.
Windows builds aren't code-signed yet, so SmartScreen will warn you.
Phones: no built-in voice yet.

Verdict

If you edit for clients with deadlines, not yet. If you're a student, a Linux user, a small creator, or a developer who wants an editor an agent can drive, it's already useful, and it's free. Nothing uploads, which also makes it a good fit for private footage.

Would you let an AI agent cut your videos through MCP? I'm curious what you'd automate first.

Links

Concat website + download: https://concat.standalones.org
Source code: https://github.com/jub0t/Concat
MCP setup: https://github.com/jub0t/Concat/blob/main/docs/MCP.md
Full video: https://youtu.be/KWJw3RQHElU

AI-assisted disclosure: I used an AI assistant to help structure this write-up. The compression test, timing and network log are my own (Chrome on my Mac, November 6, 2026). Feature details come from Concat's site, docs and GitHub repo. Not sponsored by Concat or CapCut.

I Asked Kimi K3.1 Pro to Build a 3D Game From One Prompt. It Worked. Here's the Catch

RainyTech — Fri, 30 Oct 2026 15:19:21 +0000

AI-assisted disclosure: I used an AI assistant to help structure this write-up. The test runs, timings and logs are my own (kimi.ai free plan, October 28, 2026). Benchmark numbers come from Moonshot's blog and Artificial Analysis, linked at the end.

Short version: one prompt to Kimi K3.1 Pro gave me a playable 3D racing game: a single 752-line HTML file using Three.js, 3 laps, 60 fps, zero page errors. But the file took about 21 minutes to arrive, Kimi's own self-test was still stuck after 38 minutes, and the free plan refused my next prompt 6 times in a row.

▶ Full video: https://youtu.be/iAoYAYFxDQs

What is Kimi K3.1 Pro?

Kimi K3.1 Pro is the new flagship model from Moonshot AI, the Chinese lab behind Kimi. It launched on October 23, 2026.

Base model: Kimi K3, about 2.8 trillion parameters (mixture of experts)
Context window: 1 million tokens
Open weights: Moonshot says the full weights will be released by November 4, 2026
API price: $2 input / $10 output per million tokens ($0.20 for cached input). A Highspeed variant costs $4 / $20.

Moonshot's headline claim is that it codes almost as well as Claude Opus 5.5. I wanted to see what a normal person gets on the free website.

Test 1: one prompt, one 3D game

This is the exact prompt I sent on kimi.ai (free plan, Kimi K3.1 Pro):

Build a playable 3D racing game in ONE single HTML file using Three.js from a CDN.
Low-poly neon city track at night, the car drives with WASD or arrow keys, chase
camera, 3 laps with a lap timer and best-lap display, a few obstacles to dodge,
and a start screen with "Press Space". No other external assets. Give me the full file.

Kimi came back with "Neon Velocity": one index.html, 752 lines, 47 KB, Three.js loaded from a CDN, a night circuit, nitro, a minimap, lap timers and a finish screen.

To check it honestly, I didn't just eyeball it. I drove it with a script that sends real key presses in headless Chrome on a Mac mini M4:

title screen ready            0.16 s
lap 1 done                    19.533 s
lap 2 done                    41.033 s
race finished                 64.750 s (shown as 1:04.749)
best lap                      0:19.516
fps                           avg 60.5, min 60
page errors                   0
draw calls / triangles        36 / 25,554

The only console message was a Three.js warning about the old non-module build. For a single prompt, that is a genuinely good result.

Test 2: speed

This is where it hurt.

The file was written about 21 minutes after I hit send.
Kimi then tried to test the game itself in its own browser. It hit a 404 on its own server, switched ports, and its step "Loads game and captures title screen" was still running at 38:53. I stopped it and took the file from the chat.

Artificial Analysis measured the API at about 42.7 tokens per second (No. 184 of 192 models) when I checked, versus about 98 for Claude Opus 5.5.

Test 3: the free plan

Next I asked for a second round: 3 AI rival cars and a race position display. The free plan refused it 6 out of 6 times, including in a brand-new chat and at a lower effort setting:

"Too many people are chatting with Kimi; a subscription will grant you priority access."

So "free" works, but not reliably during busy hours. On the API side it is cheap: Artificial Analysis lists about $3.68 to run its index tasks on Kimi K3.1 Pro versus $10.04 on Opus 5.5.

Test 4: is it really Opus-level?

Source	Kimi K3.1 Pro	Claude Opus 5.5
Moonshot (self-reported): Terminal-Bench 4.0	65.7%	66.4%
Moonshot (self-reported): BenchCAD	92.7%	96.2%
Artificial Analysis: Intelligence Index	46 (No. 20 of 51)	58

On Moonshot's own chart it's less than a point behind. The independent index puts it well behind, in 20th place. Both can be true: it is very strong at the agentic coding tasks Moonshot picked, and less strong across the board.

The catch

Slow. 21 minutes for one file, plus a self-test that never finished.
The free plan gets busy. 6 of 6 follow-up prompts refused.
"Opus-level" is Moonshot's claim, not the independent ranking.

Verdict

If you want the best coding model today, Claude Opus 5.5 is still smarter and more than twice as fast. But one free prompt built a real, playable 3D game, and the API is about a third of the price. If the open weights really land by November 4, anyone with enough GPUs can run a near-frontier model themselves. That's the part I'll be watching.

Have you tried Kimi K3.1 Pro for coding yet? I'd like to know if you hit the same free-plan wall.

Sources

Moonshot blog: https://www.kimi.ai/blog/kimi-k3-1-pro
Kimi API pricing: https://platform.kimi.ai/docs/pricing/chat
Artificial Analysis: https://artificialanalysis.ai/models/kimi-k3-1-pro
Moonshot AI on GitHub: https://github.com/MoonshotAI
Full video: https://youtu.be/iAoYAYFxDQs

MacBook Pro M6 for Developers: What the New Chip Means for Xcode, Local AI and Your Next Laptop

RainyTech — Wed, 28 Oct 2026 13:18:20 +0000

AI-assisted disclosure: I used an AI assistant to help structure this write-up. Every number comes from Apple's official newsroom post and tech specs (linked at the end). I have not tested an M6 MacBook Pro yet, so treat the speed claims as Apple's until real benchmarks land.

Short version: the 14-inch MacBook Pro with M6 starts at $1,999 (education $1,899), ships November 6, and brings a 12-core CPU, a 12-core GPU, a Dual 16-core Neural Engine, up to 32GB of memory at up to 170GB/s, Wi-Fi 7 and up to 24 hours of battery. For developers, the parts that matter most are the extra CPU cores, the memory bandwidth for local AI, and the 1TB base SSD.

📺 Full video (8 min): https://youtu.be/gnDJOewWAOg

What Apple announced

Apple announced the new 14-inch MacBook Pro with M6 on October 27, 2026, with a press release rather than a keynote. Only the 14-inch model gets M6. The 16-inch MacBook Pro stays on M5 Pro and M5 Max.

Price: from $1,999 (education from $1,899)
Pre-order: October 28 · In stores: Friday, November 6
Colors: space black and silver
Software: ships with macOS 27, including Siri AI

The M6 chip, in developer terms

Part	M6 (14-inch MacBook Pro)	What it means for you
CPU	12-core (2 super + 4 performance + 6 efficiency)	More cores for parallel builds, test runners and containers
GPU	12-core, up to 30% faster graphics (Apple)	Faster Metal work, previews and GPU-accelerated ML
Neural Engine	Dual 16-core, up to 1.9x faster (Apple)	On-device AI features and Core ML models run faster
Memory	16GB–32GB, up to 170GB/s (24GB and up)	Bandwidth is the bottleneck for local LLM token speed
Storage	1TB base SSD	Room for Xcode, simulators, Docker images and model weights

Apple also says M6 has the world's fastest CPU core. Single-core speed is what you feel in an IDE: indexing, type checking, and incremental builds that only touch a few files.

Local AI: memory is the number to watch

If you run local models with Ollama, MLX or llama.cpp, two numbers decide your experience:

How much memory you have. It sets the biggest model you can load. 16GB is tight for anything beyond small models; 24GB or 32GB gives you real headroom.
Memory bandwidth. Token generation speed is mostly limited by how fast the chip can read the model's weights. Apple lists up to 170GB/s on M6 configurations with 24GB or more.

Apple claims up to 25% faster AI workloads overall. My advice: if local AI is part of your workflow, buy 24GB at minimum, because the base 16GB configuration also has lower bandwidth.

Display, battery and ports

14.2" Liquid Retina XDR, up to 1600 nits peak HDR
Up to 24 hours of battery (Apple)
Thunderbolt 4, Wi-Fi 7 and Bluetooth 6
3.4 lb

Wi-Fi 7 is a quiet upgrade that matters if you pull large images or model weights over a fast home or office network.

The catch

It is a spec-sheet launch. Every speed number above is Apple's. Wait for independent benchmarks before you trust "30%" for your workload.
The design did not change. Same size, same weight, same display brightness, same ports class. If you want a new form factor, this is not it.
32GB is the ceiling. If you need more memory for large local models, look at the 16-inch MacBook Pro with M5 Pro or M5 Max instead.

Who should buy it

Buy: developers on an Intel Mac or an M1-era machine. The jump in CPU cores, Neural Engine and memory bandwidth is large, and the 1TB base SSD removes the usual storage upsell.
Consider: anyone who runs local AI daily, as long as you choose 24GB or 32GB.
Wait or skip: if you already have a recent Apple silicon MacBook Pro and it is not slowing you down. Wait for real benchmarks first.

Sources

Apple Newsroom: https://www.apple.com/newsroom/2026/10/apple-introduces-the-new-14-inch-macbook-pro-with-m6/
MacBook Pro: https://www.apple.com/macbook-pro/
Tech specs: https://www.apple.com/macbook-pro/specs/
Full video breakdown: https://youtu.be/gnDJOewWAOg

PHP Got Hacked? PHP2Shell, LiteLLM, GitLab and a CA Key of 4: What Devs Should Update Now

RainyTech — Sun, 25 Oct 2026 02:04:26 +0000

AI-assisted disclosure: I used an AI assistant to help structure this write-up. Every fact comes from public advisories, researchers and news reports (linked at the end). Nothing here was hacked or tested by me against anyone's systems.

Short version: update PHP to 8.5.12 / 8.4.27 / 8.3.36 / 8.2.35 and restart it, check every Python environment for a file called cook.pth, upgrade self-managed GitLab (even if you patched on October 12), and keep your OS, browser, Go, pip and certifi up to date.

October 17–23, 2026 was one of the worst weeks the internet has had in years. The NSA told everyone running PHP to get ready for an emergency update. A Python package with about 23 million weekly downloads shipped with a stealer inside. GitLab patched a maximum-severity bug that had been hiding since 2014. Banks found AI agents wandering around inside customer accounts. And a company trusted to protect HTTPS used a secret key that was... the number 4.

📺 Full video (8 min): https://youtu.be/KNXtuCbvxDw

1. PHP2Shell: a 9.2 bug in PHP itself

PHP runs roughly 70% of websites, including WordPress, Drupal and Joomla. PHP2Shell is a pair of flaws in PHP 5.4 through 8.5:

CVE-2026-109618: out-of-memory corruption that leads to unauthenticated code execution as the PHP web worker. Researchers demonstrated pre-auth RCE on WordPress, Drupal and Joomla.
CVE-2026-109619: stale global state across requests causes a use-after-free in the upload progress feature, which leaks memory or runs code.

Both are rated CVSS 9.2 (watchTowr rated it 10.0). Attackers could read passwords, tokens, API keys and certificates straight out of PHP process memory, including data from other people's requests.

Who found it: AI. Ryan Austin at Calif found it in an August 2026 audit, part of OpenAI's "Patch the Planet" project, using GPT-5.6 Sol. Wiz Research found it independently. A PHP core developer had spotted the out-of-memory error back in a 2024 audit, and it was never fixed.

How it got out early: details leaked on Hacker News on October 20, just under two days before the patch. Amazon Threat Intelligence saw the first exploit attempt on October 17. Cloudflare's CEO said they saw thousands of attempts before the announcement and millions after. Dutch, Finnish and French government sites went offline, and the NSA advised sites that couldn't update to go offline too.

What to do:

php -v
# fixed: 8.5.12, 8.4.27, 8.3.36, 8.2.35 (source-only backports for 7.4, 8.0, 8.1)

Then restart Apache, PHP-FPM and any long-running workers so the old code is actually gone from memory. There is no complete workaround: upgrading PHP is the only fix.

2. LiteLLM 1.107.0: a backdoored Python package

LiteLLM is a Python proxy library for LLM providers with about 23 million weekly downloads. On October 21, a malicious 1.107.0 was live on PyPI from 11:39 to 13:20 UTC (Docker images too). Wiz counted about 49,000 downloads in those two hours.

The attackers used stolen GitHub CLI OAuth tokens from three engineers to run the release workflow against a lookalike fork, Berri-AI/litellm (the real org is BerriAI).

The clever, nasty part: it dropped three files into site-packages:

cook.pth
cook.py
sitecustomize.py

A .pth file in site-packages runs every time Python starts. So the stealer ran even if you never imported litellm, and pip uninstall litellm does not remove these files. It went after browser passwords and cookies, AWS/GCP/Azure credentials, OpenAI/Anthropic/Cohere API keys, SSH and GPG keys, and shell history, and sent them to the lookalike domain litellm[.]dev. About 4.5 hours later the attacker was already testing stolen AWS keys for Bedrock access.

Check every Python environment you have:

python -c "import site; print(site.getsitepackages())"
# look in each path printed above:
ls <site-packages-path> | grep -E '^(cook\.pth|cook\.py|sitecustomize\.py)$'
pip show litellm | grep Version

If you find them: delete all three files, go back to 1.106.2, and rotate every credential that machine could see, from a clean machine.

3. GitLab CVE-2026-110264: a 10.0 going back to 2014

An integer overflow while compiling a crafted regular expression lets an unauthenticated user run code on a self-managed GitLab server. CVSS 10.0. Affected versions go all the way back to GitLab 7.0, released in 2014. Censys counted 81,820 internet-visible GitLab hosts.

Fixed releases: 19.5.1, 19.4.3, 19.3.6, 19.2.9, 19.1.9, 19.0.10, 18.11.13. No workaround.

The trap: if you patched on October 12, you're still vulnerable. That was CVE-2026-97229, a different 10.0 regex bug from the same researcher. GitLab.com and GitLab Dedicated are already patched.

4. AI agents inside bank accounts

Axios reported on October 22 that third-party AI agents were logging into US bank customer accounts, sometimes with the customers' own usernames and passwords, and some took actions that "appear to go beyond what anyone asked them to do." Banks often can't tell whether a login is a person or a machine.

Morgan Stanley is blocking outside agents; JPMorgan, Bank of America and Citi are restricting them. (Some of this reporting comes from anonymous sources.)

The dev lesson: if you build agents, use proper delegated, scoped access (OAuth, read-only where possible). Never ask users for raw bank credentials. As a user: never hand your bank password to an AI agent.

5. eMudhra: the private key was 4

eMudhra is an Indian certificate authority. Eight of its intermediate CA private keys could be computed straight from the public certificates. The keys were 4, 56, 66, 115, 155, 156, 230 and 237. A P-384 private key should be a random 384-bit number (about 3.9 × 10¹¹⁵ possible values). These all fit in one byte. Scott Helme recovered all eight on a laptop in under a second.

eMudhra said a flaw in its April 2025 key generation ceremony produced a second set of weak keys alongside the strong ones. Chrome blocked them on October 17 about 100 minutes after the report, and Firefox, Apple and Microsoft followed. No fraudulent certificates were found in Certificate Transparency logs, but anyone with those keys could have signed a certificate for any domain.

What to do: update your OS, browser and runtimes. Go shipped fixes in 1.27.3 and 1.26.10 (CVE-2026-111943), and certifi, pip, uv and Electron all got patches. If you ship a Go binary or a bundled CA store, rebuild it.

Your checklist

PHP → 8.5.12 / 8.4.27 / 8.3.36 / 8.2.35, then restart PHP-FPM / Apache / workers
Search every venv, container and CI image for cook.pth; LiteLLM → 1.106.2; rotate keys
Self-managed GitLab → 19.5.1 / 19.4.3 / 19.3.6 / 19.2.9 / 19.1.9 / 19.0.10 / 18.11.13
Go → 1.27.3 or 1.26.10; update certifi, pip, uv, Electron; rebuild bundled CA stores
No raw bank credentials in any AI agent you build or use

Which one hit you hardest this week? Tell me in the comments.

📺 Full video: https://youtu.be/KNXtuCbvxDw

Sources

PHP2Shell: https://www.php2shell.com/ · https://en.wikipedia.org/wiki/PHP2Shell · https://www.php.net/
CNBC: https://www.cnbc.com/2026/10/22/php2shell-vulnerability-what-is-it-and-is-your-data-at-risk.html
LiteLLM incident report: https://docs.litellm.ai/blog/security-incident-oct-21-2026
Wiz Research: https://www.wiz.io/blog/litellm-compromised-steals-credentials
Datadog Security Labs: https://securitylabs.datadoghq.com/articles/litellm-compromised-pypi-docker/
BleepingComputer (GitLab): https://www.bleepingcomputer.com/news/security/gitlab-warns-of-new-max-severity-code-execution-vulnerability/
Axios on X: https://x.com/axios/status/2113198996759957511
Google Security Blog (eMudhra): https://blog.google/security/protecting-chrome-and-android-users-from-compromised-emudhra-intermediate-ca-certificates/
The Next Web (eMudhra): https://thenextweb.com/news/emudhra-certificates-weak-keys-browsers-block

iPhone Duo Is Out: Which Storage Should You Buy? 256GB vs 512GB vs 1TB vs 2TB

RainyTech — Sat, 24 Oct 2026 03:32:07 +0000

AI-assisted disclosure: I used an AI assistant to help structure this write-up. Prices and specs come from Apple's iPhone Duo page, tech specs and Newsroom (linked at the end). Not sponsored by Apple.

Short answer: most people should buy the 256GB iPhone Duo. If you film a lot of 4K or keep phones for years, get the 512GB. The 1TB is for working creators, and the 2TB is money almost nobody will use.

The iPhone Duo has been in stores since Friday, October 23, 2026. The base model is $1,999, but the one you see in every ad can cost $3,199. That's a $1,200 gap for the exact same phone: same screen, same chip, same cameras. You only pay more for storage.

📺 Full video (8:04): https://youtu.be/0DHT--R3RiI

The four prices

256GB: $1,999
512GB: $2,199 (+$200)
1TB: $2,599 (+$400)
2TB: $3,199 (+$600)

Notice how the steps grow. Night Sky and Star White cost the same. Monthly payments start at about $83.29/month over 24 months.

What every iPhone Duo gets (storage changes none of this)

7.6-inch inner display, which Apple says is 50% bigger than the iPhone 18 Pro Max, plus a 5.4-inch outer display
5.2 mm thin when open, the thinnest iPhone Apple has made
A20 Pro chip (same as iPhone 18 Pro)
Two 48MP Fusion cameras, 4K at up to 120 fps in Dolby Vision
Up to 44 hours of video playback on the outer display
Titanium frame and hinge cover, IP68
About 50% charge in ~20 minutes with a strong wired charger, MagSafe up to 25W
USB-C up to 10 Gb/s

The cheapest iPhone Duo is just as fast as the most expensive one.

How much storage do you actually need?

iOS and built-in apps take about 12–24 GB, so the 256GB model leaves roughly 230 GB for you.
The camera is where storage disappears. 4K video takes hundreds of megabytes per minute even in Apple's efficient format, and a few long 4K clips from a trip can fill dozens of gigabytes.
There's no memory card slot. Whatever you buy, you live with for years.
Games and offline movies add up, especially on a 7.6-inch screen.
iCloud can keep full-size photos online and smaller copies on the phone. That alone saves many people from buying more storage, but it's a monthly bill and needs a good connection.

Quick check: Settings → General → iPhone Storage. If you use less than 150 GB today, the 256GB model is enough.

My pick for each buyer

256GB ($1,999): most people. Browsing, messaging, video, normal photos, especially if you already pay for iCloud.
512GB ($2,199): the sweet spot. Double the space for $200, the cheapest gigabytes in the lineup. Pick it if you film a lot of 4K, play big games, skip iCloud, or keep phones 4–5 years.
1TB ($2,599): creators who record long 4K every week or edit on the phone.
2TB ($3,199): skip it. $1,200 over the base model is almost a second phone. With USB-C at 10 Gb/s, a cheap external drive beats paying $600 more.

The catch

No Face ID. You unlock with Touch ID in the side button.
No telephoto lens. Zoom tops out at 2x optical quality. The iPhone 18 Pro has a real 48MP telephoto for $800 less.
eSIM only. No physical SIM.
Apple Pencil support comes later this year, not at launch.
Siri AI is in beta, English first, not in the EU at launch.
It's Apple's first-generation foldable. If you're unsure, wait for real long-term reviews.

If you love zooming at concerts or sports, want Face ID, or want the best value, the iPhone 18 Pro (from $1,199) is the better buy. If you want a tablet-sized screen that folds into your pocket, nothing else from Apple does that.

How to pay less

Trade-in credit up to $885 (iPhone 13 or newer)
Carrier credits at Apple up to $1,200 (AT&T)
Apple Upgrade lease from about $57.99/month. Compare the final price, not just the monthly one.

Final answer

256GB for most people. 512GB if you film a lot or keep phones for years. 1TB only for pro creators. 2TB is money you'll almost certainly never use.

Which one are you getting? Tell me in the comments.

📺 Full video: https://youtu.be/0DHT--R3RiI

Sources

Apple, iPhone Duo: https://www.apple.com/iphone-duo/
Apple, iPhone Duo tech specs: https://www.apple.com/iphone-duo/specs/
Apple Newsroom: https://www.apple.com/newsroom/2026/09/apple-unveils-iphone-duo/

How an AI Agent Taught NVIDIA's $399 Robot Duck to Walk: Prompt to Policy in 7 Steps

RainyTech — Sat, 24 Oct 2026 03:31:44 +0000

AI-assisted disclosure: I used an AI assistant to help structure this write-up. This is **not* a hands-on test. Everything here comes from NVIDIA (the keynote and its RTX Spark demo), Pollen Robotics / Hugging Face, and news coverage linked at the end.*

At NVIDIA's GTC Berlin keynote on October 21, 2026, Jensen Huang gave four tiny robots one command: "Take a nap." Three dropped instantly. The fourth kept walking for a few steps before tipping over.

The fall is funny, but it isn't the story. The story is how they learned to walk at all: thousands of virtual copies trained in a world that an AI agent built on a laptop.

📺 Full video (8:00): https://youtu.be/R1VtPIjb6Es

What a Microduck is

A tiny walking robot from Pollen Robotics, the robotics team Hugging Face bought in April 2025.
About 25 cm (10 in) tall, under 2 lb.
15 motors (legs, neck, head), a front camera, a depth sensor and two IMUs.
Walks, sits, crouches, catches its balance when pushed, and gets back up on its own. Seven behaviours come trained out of the box.
$399 pre-order (opened Aug 27), now sold out. First shipments expected by Christmas in North America and Europe.
The training code is open source, so owners can teach it new tricks.

The pipeline: prompt to walking robot in 7 steps

NVIDIA's demo ran this stack on an RTX Spark laptop (N1X chip, "one petaflop" per Jensen), bursting big jobs to the cloud:

Agent harness: Hermes
Models: Nemotron and OpenAI's GPT-6 Astra
Tools: FreeCAD, Blender, Omniverse SimReady, Isaac Sim, the Newton physics engine, Isaac Lab

The steps:

Prompt. Describe the course and the skill: "plan an obstacle course to teach Microduck how to walk."
Plan. Hermes + Nemotron lay out the course as a task-by-task workflow, listing every object, and even generate a concept image first.
Build. GPT-6 Astra models each 3D object in FreeCAD, applies materials in Blender, and adds real physical properties from Omniverse SimReady libraries.
Simulate. Astra composes the course in Isaac Sim, rigs the duck's joints, actuators and links, and simulates movement with Newton.
Train. In the cloud, thousands of Microducks train in parallel in Isaac Lab with reinforcement learning. Each virtual duck tries a move, gets a score, and the policy is adjusted. Omniverse also simulates the cameras, so the duck learns to see and avoid obstacles.
Evaluate. The policy comes back to the RTX Spark, where Nemotron and Isaac Sim test it with many robots running at once.
Deploy. The validated policy goes onto the real duck.

Jensen's line from the keynote sums up the approach: "The way you build an agent is you ask an agent to build it for you."

Why sim-to-real matters

Pollen shows the same idea in its own footage: a move learned in simulation runs on the real robot, and when a move fails in simulation, it fails the same way on the real robot. That's the whole point. You crash the virtual duck a thousand times, not the real one.

The catch

The nap is the honest part of the demo. A simulation is never exactly the real world. Real floors, lights and noise differ, so robots still slip up.
The laptop didn't do the heavy lifting. Training ran on NVIDIA supercomputers in the cloud. The RTX Spark planned the work and checked the result.

Why developers should care

This is NVIDIA's "physical AI" pitch in miniature: an agent builds the world, the cloud trains the policy, a local machine evaluates it, and the robot runs it. The same keynote announced a 31-degree-of-freedom reference humanoid (Agile ONE SN) built on Isaac GR00T, and Reuters reported Agile Robots plans to produce 1,000 to 2,000 small humanoids next year with NVIDIA.

A $399 open-source duck is the cheapest way to see that recipe end to end.

Would you buy one and teach it new tricks? Tell me in the comments.

📺 Full video: https://youtu.be/R1VtPIjb6Es

Sources

NVIDIA GTC Berlin 2026 keynote: https://www.youtube.com/watch?v=PvFjbPYEOZI
NVIDIA, "From Prompt to Walking Robot With NVIDIA RTX Spark": https://www.youtube.com/watch?v=VvJai6HGfyE
NVIDIA blog, keynote live updates: https://blogs.nvidia.com/blog/gtc-berlin-2026-keynote/
Pollen Robotics, "Meet Microduck": https://www.youtube.com/watch?v=reiTh7K4KSc
Pollen Robotics, sim to real: https://www.youtube.com/watch?v=szW7N_7B3tU
Reuters (via KSL): https://www.ksl.com/article/51639111/nvidias-huang-hails-europes-ai-potential-flanked-by-robot-ducks
The Register: https://www.theregister.com/ai-and-ml/2026/10/21/jensen-huang-sends-in-the-robot-ducks-to-sell-nvidias-ai-everywhere-vision/5304472

Is FilmCraft 100% Free? I Scored the Open-Source Premiere Pro Clone in 10 Rounds

RainyTech — Sat, 24 Oct 2026 02:57:56 +0000

AI-assisted disclosure: I used an AI assistant to help structure this write-up. Every number comes from my own runs on my Mac or from the sources linked at the end.

Short answer: yes, FilmCraft is free. It's open source (MIT OR Apache-2.0), it rebuilds Adobe Premiere Pro from scratch in pure Rust, and it costs $0. Premiere Pro costs US$22.99/month on the annual plan.

But is a free Premiere clone actually usable? I gave FilmCraft 0.9.0 real jobs on a Mac mini M4 (16 GB) and scored it against Premiere in 10 rounds.

📺 Full video (5:38): https://youtu.be/0moxAPMMego

What FilmCraft is

A clean-room reimplementation of Premiere Pro: same panels, workspaces, tools and keyboard shortcuts (with shortcut presets for Premiere, Final Cut Pro and Avid).
Pure Rust, no FFmpeg inside. It ships its own H.264, HEVC, ProRes and AAC codecs.
From the ArtCraft team. The repo was created Sep 30, 2026 and had 14,962 stars and 6,249 forks 19 days later.
Runs on macOS, Windows, Linux and in the browser (WebAssembly).
650+ engine commands, a CLI and an MCP server, so an AI agent can drive the editor. This is the part developers should care about most.

The scorecard

Round	Winner	Score (FilmCraft–Premiere)
1. Interface & shortcuts	Tie	1–1
2. Editing & trimming	Tie	2–2
3. Green-screen keying	Tie	3–3
4. Effects & transitions	Premiere	3–4
5. Color grading	Tie	4–5
6. Audio	Premiere	4–6
7. Captions & AI	Premiere	4–7
8. Formats & export	Premiere	4–8
9. Projects & plugins	Premiere	4–9
10. Price & freedom	FilmCraft	5–9

Premiere wins on depth. FilmCraft wins on price.

Honesty note: Premiere was not installed on my Mac. The Premiere side of each round uses Adobe's own published docs and product pages. The FilmCraft side is real runs.

The developer angle: a whole edit from one script

"Every menu item, button, slider and drag in FilmCraft is a command" (that's from the README). So I built a 15-second 4K coffee-shop ad without touching the timeline:

a new sequence, 5 clip placements, 4 cross dissolves
a Lumetri grade on all 5 clips (9 settings each)
2 titles

That's 58 commands, run from one script in 0.01 s. The same command surface is exposed through the CLI and an MCP server, which means an LLM agent can assemble, grade and export a video. Premiere can't do this without a plugin SDK and a lot of glue.

Real numbers from my Mac mini M4

Test	Result
Build the 15 s 4K ad (58 commands)	0.01 s
One 4K frame preview	~0.6 s
Export 15 s 4K H.264, software	82 s (4.4 fps)
Export 15 s 4K H.264, hardware encoding on	131 s (2.8 fps), slower
Export 13 s 1080p30 meme, H.264	27.4 s
Export to FCP7 XML (what Premiere imports)	valid: 7 clip items, 4 transitions

Yes, hardware encoding was slower than software. Lumetri, keys and export rendering still run on the CPU in this build.

Green-screen keying: a 1:1 copy, with a catch

FilmCraft's Ultra Key has the same 18 parameters with the same defaults as Premiere's (Transparency 45, Highlight 10, Shadow 50, Tolerance 50, Pedestal 10, and so on).

But on my green-screen clip (a tourist with a map, backdrop rgb 102,221,9) the default key colour didn't key it at all. Setting the key colour to the backdrop and switching to "Aggressive" keyed it cleanly. Every meme in the video was made this way.

Where it falls short

Can't open .prproj files. This is the team's main beta blocker. Importing a Premiere sequence via XML took 97 s in their own report.
No plugins: no VST3 / Audio Units, no OpenFX, no Premiere/After Effects plug-in API.
Auto captions don't work out of the box on Mac. Offline Whisper transcription exists, but the macOS release build says model downloads aren't available ("built without speech-download").
H.264 is 8-bit 4:2:0 only, no camera RAW, and no AV1 export.
The team's own parity report: ~86% of Premiere's features, but only ~56% "ready for real work". Status: alpha.

Verdict

Pro client work? Stay on Premiere.
Memes, shorts, school projects, small ads? FilmCraft works, for $0.
Building an AI video pipeline? FilmCraft's command/CLI/MCP surface is the most interesting thing here, and it's open source, so you can read exactly what each command does.

Would you switch? Tell me in the comments.

📺 Full video with all 10 rounds: https://youtu.be/0moxAPMMego

Sources

FilmCraft repo and README: https://github.com/storytold/filmcraft
FilmCraft parity report (vs Premiere Pro 26.5.2): https://github.com/storytold/filmcraft/blob/main/docs/target-app-parity.md
Adobe Premiere Pro pricing: https://www.adobe.com/products/premiere.html
Adobe Help Center, "Ultra Key effect parameters" (updated Jan 7, 2026)

PhotoCraft Review: I Tested the Free, Open-Source Photoshop Alternative With 4 Real Edits

RainyTech — Sun, 18 Oct 2026 12:27:24 +0000

Photoshop costs $22.99 a month. PhotoCraft costs $0, it's open source, and in under three weeks it passed 66,000 stars on GitHub.

So I put it to work on my Mac mini M4 with four real edits, the kind most of us actually do: cutting someone out, a caption meme, fixing a dull photo, and text for a meme video. Here is what worked, step by step, and what it still can't do.

▶ Full video: https://youtu.be/O4necNCA5D4

What is PhotoCraft?

PhotoCraft is a free copy of Photoshop, rebuilt from scratch in the Rust language. It's part of ArtCraft, a family of open-source clones of creative apps.

Open source: all the code is public on GitHub, so anyone can read it, fix it or improve it.
Runs everywhere: Mac, Windows, Linux, and even in your web browser.
Offline: no account, no subscription, no cloud.
Opens real PSD files with your layers still editable.

Install and quick tour

Go to getartcraft.com (or the GitHub releases page) and pick the file for your computer.
On a Mac, open the DMG and drag PhotoCraft into Applications.
Open it. If you've used Photoshop, it will feel familiar: 67 tools on the left (move, select, crop, brush, text…) and layers on the right. Every edit sits on its own layer, like sheets of glass.

Edit 1: Cut out anyone (The Rock eyebrow meme)

Open a frame from the meme.
Press Command + F and search for Subject. PhotoCraft finds the person by itself, in about a tenth of a second.
Layer → Layer Mask → Reveal Selection. The background is gone.
Open any background photo, Select All → Copy, and paste it into the meme.
Layer → Arrange → Send to Back, so it sits behind him.
Add a caption with the Type tool, then give it a Stroke from the fx button so it pops on any background.

Result: a brand-new meme, and every layer is still editable.

Edit 2: The classic caption meme (Surprised Pikachu)

Open a frame from the meme video.
Image → Canvas Size. Anchor at the bottom, make it taller, and fill with white. Now there's room on top for the joke.
Type tool, black text, write the caption. Tip: the best captions fit in two lines.
File → Export as a PNG, and post it.

Edit 3: Make a dull photo pop, the pro way

Layer → New Adjustment Layer → Curves. Bend the line into an S shape for more contrast.
Add a Vibrance adjustment layer to bring the colours back.

Adjustment layers never change your original photo: turn them off and it's untouched. The same two layers work on selfies, product photos and thumbnails.

Edit 4: Text for a meme video

File → New, background Transparent.
Write your caption and give it a stroke and a drop shadow.
Export as PNG. The empty parts stay see-through.
Drop that PNG on top of any clip in your video editor.

Bonus: scripts and AI agents can drive it

Every button in PhotoCraft is also a command, over 500 of them. So scripts, and even AI agents, can run the same edits, with the same engine and the same results.

The honest part

PhotoCraft is still early alpha. The team says it's not yet a Photoshop replacement for daily professional work.
There's no generative AI yet, some tools are missing, and you will hit bugs.
But for memes, thumbnails and quick photo fixes, it already does the job, for free.

Price check (October 18, 2026)

Adobe Photoshop: US$22.99/month (annual plan, billed monthly, adobe.com)
PhotoCraft: free, open source