What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
TSK-1 is Taskade’s benchmark for judging whether an AI model can turn one fixed request into a complete, usable application inside Taskade Genesis. Instead of scoring only generated text or code fragments, it evaluates the product a customer can open, use, edit and query. Each model receives the same request, produces one to three builds per test, and is assessed on interface quality, task fidelity, data reliability and its ability to handle a follow-up change. The resulting tier scores are converted to a 0–100 convenience scale. Because most comparisons rely on very small samples and software versions change, TSK-1 results are dated directional evidence rather than permanent rankings.
What is the TSK-1 methodology?
TSK-1 is a hands-on, customer-style evaluation created by Taskade for AI app building. The benchmark’s unit of grading is the finished product, not the model’s prose, hidden chain of thought or an isolated code sample. As Taskade’s 2026 methodology puts it: “The unit of grading is what a customer could do with that product on the day it was tested.”
As an Amazon Associate I earn from qualifying purchases.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minutePC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11A model must therefore produce an app with working pages, correctly typed database fields, automations and a grounded built-in AI assistant. A visually attractive mock-up that loses records, calculates a score incorrectly or fails after an edit can score below a less flashy but dependable application.
The method is designed around a single practical question: can a person give an AI builder a realistic brief and receive a product they can use immediately, then safely modify?
Popular on Amazon
- Read Before You Buy — No Video Output: These adapters support charging and USB 2.0 data transfer, but cannot transmit video signals. Except for standard USB webcams (which use USB data only), they are not compatible with HDMI/DisplayPort cables, video-capable USB-C hubs, or docking stations with video output.
- Convert USB-A Ports to USB-C: Designed to connect USB-C earphones, cables, flash drives, card readers, and other USB-C accessories to standard USB-A ports. Plug-and-play with no drivers or software required.
- Aluminum Alloy Housing: Built with a sturdy aluminum alloy shell that aids in heat dissipation and protects against daily wear and scratches. Designed to maintain a stable and secure connection.
- Compact & Travel-Friendly: The ultra-compact design allows the adapter to stay plugged into your device without blocking adjacent ports or adding bulk, reducing wear and tear on your original USB ports.
- 12-Month Warranty: Backed by a 12-month manufacturer warranty for peace of mind. Designed to meet strict quality control standards for reliable everyday performance.
How are TSK-1 tests controlled?
Comparisons are controlled by freezing the request and the build conditions. Taskade fingerprints the request text before scored builds, so even a small wording change creates a different test. Two results are directly comparable only when the request bytes match.
- Same brief: every compared model receives identical request text, including formatting and requirements.
- Same environment: the Taskade Genesis version, instructions and other stated conditions remain constant within a comparison.
- Complete record: every build is retained, including incomplete and failed attempts, rather than showing only the best result.
- Date attached: claims are tied to the model and builder versions available at the time.
Results from different weeks are directional, not laboratory-equivalent. Models can change, Genesis can change, and the same request can behave differently after an update. A score should therefore be read with its test date, model/version name and build count.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →What requests does the benchmark use?
The published method describes two anchor requests that deliberately combine interface, data, calculations and workflow requirements.
Dota 2 match tracker
This brief asks for hero, result, KDA, match duration and notes, plus win-rate and streak dashboards. It also specifies a premium esports HUD visual style. The request tests whether a model can build both the underlying record structure and a dashboard that presents derived statistics correctly.
Customer sign-up and eligibility workflow
The second anchor is based on a real customer’s 32-question client sign-up form. It requires a scoring formula, automation and an eligibility dashboard. The complete customer form is not reproduced publicly, so outside readers cannot recreate every private question from the published description alone.
These are not toy prompts. They require an app to preserve typed inputs, calculate values, trigger actions on the right record and answer questions using newly saved information.
Also popular on Amazon
- 5-in-1 USB-C Hub: Experience comprehensive connectivity featuring a Power Delivery input, two USB-A 2.0 ports, a USB-A 3.0 port, and an HDMI port. (Note: The USB-C power delivery input port is only for connecting an external wall charger to power your laptop and cannot power peripheral devices.)
- 90W Pass-Through Charging: Achieve optimal charging with 90W pass-through power to your laptop, supported by a total input of 100W, with the hub reserving 10W for operational efficiency. (Note: Wall charger not included.)
- Quick Data Transfers: Accelerate your productivity with rapid data transfers using a high-speed 5Gbps USB 3.0 port and two 480Mbps USB 2.0 ports.
- 4K HDMI Display: Enhance your visual experience with a hub capable of delivering 4K resolution at 30Hz in both mirror and extend modes. Please note that this hub is compatible with MacBook (macOS 12 and newer), Windows 10 and 11, ChromeOS, and laptops equipped with DP Alt Mode and Power Delivery. Note: This device is not compatible with Linux.
- What You Get: Anker USB-C Hub (5-in-1, 4K HDMI), welcome guide, 18-month warranty, and our friendly customer service.
How is a generated app tested?
TSK-1 uses a fixed sequence that treats the result like a customer product rather than a screenshot. The checks are performed on the saved application produced by the model.
- Open the app in three layouts. The evaluator confirms that it launches in light mode, dark mode and a phone-width view. Since August 5, 2026, an app that does not open receives no quality score at all.
- Use a known persona. A fixed test persona with a predetermined outcome is entered so formulas and eligibility decisions can be checked against an expected result.
- Inspect every saved field. The evaluator compares each stored value with what was entered. A successful submit button is not enough if fields are missing, renamed incorrectly or stored in the wrong type.
- Recalculate derived values. Values such as KDA, win rate or an eligibility score are independently recomputed and compared with the app’s output.
- Verify automation. The test confirms that an automation ran on the correct record and did not act on an unrelated or duplicate entry.
- Question the assistant. One grounded question is sent to the built-in assistant using the newly saved record. The answer must reflect that record rather than generic text or an invented value.
- Request one real change. The evaluator asks for a specific modification, then checks that existing data remains intact and that a genuinely new page or function appears when the request requires one.
The opening gate was added after three of nine builds in an August 1, 2026 test never ran. That failure is treated as a product-level result, not silently discarded.
What do Interface, Task, Memory and Adapt measure?
Every build is graded across four qualities. Each quality describes a different way an AI-generated app can succeed or fail.
Interface
Interface covers completeness and polish across light mode, dark mode and phone width. It includes whether pages are usable, whether the visual treatment fits the request and whether responsive layouts remain coherent instead of merely shrinking desktop content.
Recommended Free Tools
Task
Task measures fidelity to the brief. Word-for-word requirements matter, as does avoiding unrequested extras that distract from the requested workflow. An app can look polished and still lose Task credit if it omits a field, changes a rule or substitutes a different process.
Memory
Memory checks whether submitted and derived data remains correct and usable after it is saved. The category includes field integrity, formulas, automation inputs and the assistant’s ability to ground an answer in the new record.
Also popular on Amazon
- Sleek 7-in-1 USB-C Hub: Features an HDMI port, two USB-A 3.0 ports, and a USB-C data port, each providing 5Gbps transfer speeds. It also includes a USB-C PD input port for charging up to 100W and dual SD and TF card slots, all in a compact design.
- Flawless 4K@60Hz Video with HDMI: Delivers exceptional clarity and smoothness with its 4K@60Hz HDMI port, making it ideal for high-definition presentations and entertainment. (Note: Only the HDMI port supports video projection; the USB-C port is for data transfer only.)
- Double Up on Efficiency: The two USB-A 3.0 ports and a USB-C port support a fast 5Gbps data rate, significantly boosting your transfer speeds and improving productivity.
- Fast and Reliable 85W Charging: Offers high-capacity, speedy charging for laptops up to 85W, so you spend less time tethered to an outlet and more time being productive.
- What You Get: Anker USB-C Hub (7-in-1), welcome guide, 18-month warranty, and our friendly customer service.
Adapt
Adapt measures the follow-up edit. The evaluator looks for a clean change that preserves existing records, adds the requested page or behavior and does not break the original workflow.
How does the TSK-1 score work?
Each of the four qualities receives one of five tiers. The tiers are converted to points and then rescaled from a possible 16 points to a 0–100 score.
| Tier | Points per quality | Meaning |
|---|---|---|
| Leading | 4 | Evidence shows the quality performing at the highest published level. |
| Strong | 3 | The quality is dependable with limited shortcomings. |
| Emerging | 2 | Useful capability is present, but important gaps remain. |
| Limited | 1 | Only partial or fragile evidence supports the quality. |
| Not scored | 0 | The build cannot provide usable evidence for that quality. |
The arithmetic is: (Interface + Task + Memory + Adapt) ÷ 16 × 100. This is a convenience scale, not 100 independent checks. Two models with the same four tiers tie, and a one-point change in one category can move the displayed score even when the underlying evidence is close.
What were the August 2026 TSK-1 results?
Taskade’s August 2026 table reported the following scores. They describe the builds and versions available at that time, not a current leaderboard.
| Model | TSK-1 score |
|---|---|
| Claude | 81 |
| GPT | 81 |
| DeepSeek | 81 |
| Kimi | 75 |
| GLM | 69 |
| Grok | 69 |
| Qwen | 50 |
| Gemini | 44 |
The equal 81 scores are a useful reminder that the scale reports tier combinations, not fine-grained performance differences. A reader comparing models should open the dated benchmark entry and inspect the evidence behind each category rather than treating the order as a permanent ranking.
Is the TSK-1 score reliable?
TSK-1 is useful for exposing practical failure modes, but its small samples limit how precisely scores can be interpreted. The method reports one to three builds per model per test. The later benchmark hub says most results rest on one build, although some experiments used five.
Also popular on Amazon
- Dual Converters, Infinite Potential：Includes 2× USB C male to USB A female adapters and 2× USB A male to USB C female adapters. Perfect for a wide range of uses—tablets with Bluetooth keyboards, expand USB ports on macbook, and more. Two different converters for all your daily needs
- Next-Level 10Gbps & 3A Charging: No more slow 480Mbps, this usb to usb c adapter has a transfer speed of up to 10Gbps, allowing you to do more transferring in less time. This usb adapter fits both USB A and USB C charger, supporting up to 3A fast charging
- Upgraded Exquisite Craftsmanship: With an aluminum alloy housing and metal connector, the usbc to usb adapter is extremely durable and sturdy. Rigorously tested to withstand more than 10,000 times of plugging and unplugging, ensuring long-lasting performance
- Widely Compatible: Transform any USB-C port into a USB-A port and USB-A port into a USB-C port, and connect up a wide range of USB-C(including MacBook, iPad Pro/Air/Mini, iPhone models with USB C ports, Samsung Galaxy series, and Google Pixel series. ) and USB A devices(including external hard drives, phones, mice, printers, and more)
- Please Note: To reach 10Gbps speed, keep the cable under 3.3 ft. For USB A Male to USB C adapters, try flipping the USB C connector. USB C Male to USB A adapters support bidirectional 10Gbps transfer within 3.3 ft
Taskade also reports noticeable run-to-run variation when the same request is executed on identical setups. A model difference should exceed that variation before being treated as meaningful. Close scores are therefore directional; they do not establish that one model is universally better for every app type.
There are concrete examples of why this matters. In the MiniMax M3 case, Taskade reported that 47.2 percent of build actions had failed. That figure describes the cited case and test conditions, not a general failure rate for every project or future version.
- Check the exact request fingerprint before comparing two entries.
- Check the builder and model versions and the test date.
- Check how many builds contributed to the result.
- Read failures and incomplete builds, not just successful screenshots.
- Look for a gap larger than the observed run-to-run variation.
What does TSK-1 not publish?
The methodology is deliberately transparent about several omissions. Taskade does not publish absolute per-model cost figures across incompatible billing periods, internal test identifiers or provider and infrastructure details. It publishes relative cost language instead.
The complete 32-question customer form is also private. Public descriptions identify its shape, scoring formula, automation and eligibility dashboard without exposing the customer’s full text. Public model and version names, dated result descriptions and the test sequence are available, but they do not provide enough information to reproduce every internal condition exactly.
Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →How can you reproduce the benchmark?
A reader can reproduce the public shape of TSK-1 in Taskade Genesis, but exact score parity requires the same request bytes, versions and hidden conditions. The benchmark hub allows a reader to submit the same request, select a model or let TSK-1 Auto decide.
- Copy the published request without rewording, reformatting or adding setup instructions.
- Record the Genesis version, model name/version, date, instructions and any other visible conditions.
- Run the planned number of builds and keep every result, including failures and incomplete apps.
- Open each app in light mode, dark mode and a phone-width viewport.
- Enter a persona with known expected outputs.
- Compare every saved field and independently recalculate each derived value.
- Check the relevant automation and ask the assistant a question grounded in the saved record.
- Submit one concrete change request, then verify data preservation and the requested new page or behavior.
- Publish dated evidence for both successful and failed builds, along with the build count.
Using a different prompt, changing the persona or judging only the first successful build may produce useful product information, but it is not a like-for-like TSK-1 reproduction.
Also popular on Amazon
- 5-in-1 Connectivity: Equipped with a 4K HDMI port, a 5 Gbps USB-C data port, two 5 Gbps USB-A ports, and a USB C 100W PD-IN port. Note: The USB C 100W PD-IN port supports only charging and does not support data transfer devices such as headphones or speakers.
- Powerful Pass-Through Charging: Supports up to 85W pass-through charging so you can power up your laptop while you use the hub. Note: Pass-through charging requires a charger (not included). Note: To achieve full power for iPad, we recommend using a 45W wall charger.
- Transfer Files in Seconds: Move files to and from your laptop at speeds of up to 5 Gbps via the USB-C and USB-A data ports. Note: The USB C 5Gbps Data port does not support video output.
- HD Display: Connect to the HDMI port to stream or mirror content to an external monitor in resolutions of up to 4K@30Hz. Note: The USB-C ports do not support video output.
- What You Get: Anker 332 USB-C Hub (5-in-1), welcome guide, our worry-free 18-month warranty, and friendly customer service.
How should you compare AI models with TSK-1?
Use the benchmark as a structured decision aid rather than a single-number contest. Start with whether the app opens, then examine the dimensions that matter to your project.
| Comparison question | Evidence to inspect |
|---|---|
| Does it produce a usable product? | Launch success, complete pages and behavior in both themes and at phone width. |
| Did it follow the brief? | Required words, fields, calculations and workflows, plus unrequested additions. |
| Can it be trusted with records? | Saved values, field types, formulas, automation targets and assistant grounding. |
| Can a customer change it safely? | Follow-up edit, preserved data and a genuinely added page or feature. |
| How strong is the evidence? | Request fingerprint, date, versions, number of builds and documented failures. |
For a production decision, repeat the same checks with your own domain-specific brief. A model that ranks well on a game tracker may not handle a regulated intake workflow, and a high score from an older Genesis version may not describe the current builder.
What is the practical verdict on TSK-1?
TSK-1 is a meaningful shift from judging AI by generated text alone: it asks whether a customer can actually use, verify and modify the resulting app. Its strongest evidence comes from the end-to-end checks for saved data, derived calculations, automation, assistant grounding and follow-up edits. Its main limitation is statistical, not conceptual: one to three builds and changing software versions make close scores uncertain. Treat the 0–100 number as a compact summary of dated tier evidence, then inspect the underlying build records before choosing a model.
Frequently Asked Questions
Can I submit my own app idea to TSK-1?
You can use the public benchmark hub to submit the published request and select a model or TSK-1 Auto. A custom idea can evaluate your own workflow, but its result should not be presented as directly comparable to the official scores because the request text is different.
Why does TSK-1 test both the AI assistant and automations?
An app can display correct values while failing when another feature consumes those values. Checking an automation and asking a grounded assistant question reveals whether the saved record is operationally usable, not merely visible on screen.
Does a higher TSK-1 score guarantee the best model for my project?
No. The score summarizes four qualities on specific briefs and dates. Your data model, compliance requirements, integrations and tolerance for follow-up edits may produce a different ranking.
Free tools Windows power users keep installed
One-click scans. No signup required.
Why are failed builds included instead of discarded?
Discarding failures would measure the builder’s best presentation rather than the probability that a customer receives a working app. Retaining incomplete and non-opening builds makes reliability problems part of the product evidence.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
