Why we rebuilt GradingPal: faster marking, worksheets to lab reports, standards for all 50 states plus AP and IB, and mastery levels backed by real work.
GradingPal 2.0 is faster, and it marks a wide range of work: worksheets, quizzes, exams, lab reports, slides and recordings as well as essays, including handwritten pages your students never had to type up. There's a library of 500 ready-made rubrics to start from. You can align to the standards you actually teach, across all 50 states and D.C. plus AP, IB, Cambridge and the UK exam boards. And your grades now build a mastery picture with the student's own work behind every result. Here's why we built it this way. For the full list of what changed, see what's new in GradingPal 2.0.
Picture the pile. Twenty-eight essays. A set of scanned maths worksheets from Period 3, half of them in pencil. Six lab reports. Thursday's quiz, still in the folder. And somewhere in a spreadsheet, the standards tracker you promised yourself you'd keep up to date this year.
That pile is what GradingPal 2.0 was built for.
We spent months asking teachers what actually slows them down. Almost nobody asked for more features. Some things came up again and again:
So that's what we worked on. 2.0 is the biggest release we've shipped, and this is the story behind the main decisions, including the things we decided not to do.
This is the part you'll feel in the first ten minutes.
GradingPal 2.0 is quicker end to end. Setting up an assignment takes about a minute. You watch the work come in and get graded live, so you're not staring at a spinner wondering if anything is happening. And when the drafts are ready, you move through them with one button: read, adjust if you want to, approve, next student.
The marking itself got better too, and not in a vague way. We put the most work into the kinds of work that used to trip it up:
The workspace helps here as much as the marking does. Grading is now a full screen: the student's work on one side, your rubric on the other, nothing else on the page. Each criterion shows the suggested level, the reason, and a quote from the student's own work. You tap a different level if you disagree, and the total updates as you go.
And home now opens with what's waiting for you: how many items need review, which class they're in, and what failed and why. It's a to-do list instead of a menu.
If you only know GradingPal for essays, this is the bit worth reading twice.
You can bring in more or less anything you set:
Maths and science are where teachers tell us the difference is biggest, and that's deliberate. Most grading tools are built for prose, so they treat a maths worksheet like an essay and a lab report like a paragraph. A wrong answer in maths has a story behind it, usually a step. A science explanation is judged by the reasoning, not the wording.
And your students don't need devices. This is worth saying plainly, because most grading tools quietly assume a one-to-one classroom. Your students can write on paper, the way they always have. You scan the class set as one PDF on the school copier and upload it.
GradingPal splits that scan into one submission per student. It reads the names off the pages and shows its working: 14 of 15 matched, one needs you. Confident matches say so, doubtful ones say "verify", and when two students could both be the match, it asks instead of guessing. If a page ends up with the wrong student, you drag it across on a page board. One button files them all and starts marking.
Messy handwriting is fine. Pencil, crossings-out, working squeezed into the margin, a diagram with labels running off the edge. This is what we hear about most from maths and science teachers, and it's why we keep putting work into paper rather than treating it as yesterday's problem.
The loop closes on paper too. Print the feedback as a report per student and hand it back in class: no logins, no email addresses, no devices at any point.
If you want the detail, the features page lists every assignment type and handwritten grading covers scanned paper properly.
This is the biggest change in 2.0.
Pick the standards your class works toward, once, when you set the class up. GradingPal suggests the likely ones for your location and subject, and then every assignment can be tagged against them.
What's in the catalogue:
Tagging is the part teachers expected to hate, so we made it a single tap. Every rubric line and every question has a "suggest standards" option. GradingPal proposes the matches, you accept or dismiss each one, and nothing sticks without you. Or name the standards when you create the assignment and it tags the whole rubric for you to check.
One thing worth knowing before you worry about jargon: you don't have to use mastery language at all. Pick your standard sets and you get plain per-standard class averages, in percentages, with no levels anywhere. A class marked to an exam board can stop right there. Tracking mastery levels is a separate switch.
Here's the change underneath the change.
Every grade you approve becomes evidence. As you mark through the term, those approved results build a picture of where each student stands on each standard: one row per standard for the whole class, with a bar showing how many students sit at each level, and a by-student grid when you want the detail.
And then the part that matters. Click any result and you see the work behind it. Not a number, not a summary. The actual graded assignment, the score, the date, and whether you adjusted it. Click a point on a student's progress chart and their essay opens.
That's the difference from a spreadsheet, and honestly it's the difference from most standards tracking. Saying a student is Approaching is easy. Showing why, at four o'clock on a Thursday with a parent sitting across from you, has always meant digging through folders. Now it's one click.
Four decisions inside this we argued about:
You choose how the level is worked out. There are five ways: every piece of evidence counts equally, recent work counts more, only the last few attempts count, the level a student lands on most often, or their best level so far. These give different answers from the same work. "Recent work counts more" is really a statement about whether a student can grow out of a bad start, and that's your call, not ours. Pick a scale too: Below / Approaching / Meets / Exceeds, Beginning / Developing / Proficient / Advanced, or Marzano 1.0–4.0, with the cut-offs yours to edit. Each result shows the maths in one sentence.
It won't guess early. Until there's enough work on a standard, the cell says "insufficient evidence (2 of 3)" instead of a level. You set the minimum. A level based on one quiz is worse than no level, because people act on it.
You can overrule it. Set a student's level yourself, with a reason like "re-assessed one-to-one after the exam". It shows as "set by teacher" everywhere, the evidence underneath stays as it was, and you can revert any time.
Reports come in two voices. One for you, with levels, percentages and evidence counts. One for families, in plain language - How Maya is doing on learning standards - with her recent work listed underneath. Export either as a PDF or a spreadsheet, for all time or just this term.
New to this way of grading? Our complete guide to standards-based grading covers the practice, and the standards mastery gradebook page shows the feature in detail.
Nobody enjoys writing a rubric at nine on a Sunday night. And a rubric written at nine on a Sunday night is rarely your best one.
So 2.0 comes with a library of 500 ready-made rubrics, reviewed by our team. Not generic templates either - these are drawn from the places teachers already trust:
Finding one takes seconds. Search across everything at once, filter by grade level, subject, framework or type of assessment, and flip on "aligned only" to see just the rubrics already tagged to standards. Preview any rubric in full - every criterion, level and point value - before you touch it.
When you use one, you get your own copy. Reword a criterion, add a level, move the points around, cut the row that doesn't apply to your class. The original stays exactly as it was for everyone else. And "use rubric" drops it straight into a new assignment, so the library is one click away from actually marking with it.
There are two other shelves next to it, and they matter as much as the 500.
Your school's shelf. When you create a rubric, you can share it with colleagues. Your department writes one argument-essay rubric, everyone marks to it, and nobody has to administer anything. For schools moving toward common assessments, this is the quiet win: consistency without a working group.
Your own shelf. Everything you've built or saved, with the ones you use most pinned to the top.
And if none of the 500 fits, start from your standards instead. Pick the standards you're teaching and GradingPal drafts a rubric aligned to them, which you then edit like any other. Library rubrics bring their standards tags with them too, so the rubric you pick on Sunday feeds the mastery picture you look at in November without you doing anything extra.
The rubric library page has the full detail, and if you're designing your own from scratch, our guide to building rubrics still holds up.
Sometimes you know what the feedback should say. You just want help getting there.
Ask GradingPal - your grading assistant within GradingPal, sits in the places where there's something to edit: while you're building a rubric, while you're writing questions and answer keys, and while you're marking one student's work. You type what you want in plain words. "Make criterion 3 stricter about citing sources, and add a level for exceptional work."
Here's the bit we care about. It doesn't make the change. It shows you the change.
You see the edits marked up where they live, plus a list of each one. Accept two, reject the third, and only the two land. There's undo afterwards. If something's waiting for you, the button says "review changes", so nothing is ever sitting there quietly changed.
This was a deliberate decision. When we asked teachers what annoyed them about AI writing tools, it wasn't bad suggestions. It was good suggestions applied without asking, which left them re-reading their own rubric to work out what had moved.
Every teacher marks a bit differently, and 2.0 finally lets you say so.
There are two lists on each assignment, and keeping them separate matters. Grading rules change how work is scored: reward what the work does well, not what it's missing, or accept g = 9.81 unless the question says otherwise. Feedback instructions change how comments are written: sound like me, not a robot, or ask a question instead of giving the answer.
Pick from a ready-made list or type your own. Mark one "always include this" and it turns up on every new assignment, for all your classes or just the ones you name. It arrives as a visible row you can remove, never slipped in behind the scenes. And you can always see the exact words that will be used, not a friendly summary of them.
Two more controls teachers ask about constantly:
Grammar check is a switch. Turn it on and spelling, punctuation and grammar findings are underlined right in the student's work, in their own sentences, with a count in its own tab. Dismiss anything you disagree with. It never moves a score. If mechanics should count, that belongs in your rubric, where it always has.
Feedback style stayed as six named styles - Targeted, Glow & Grow, Structured, Sandwich, Actionable and Socratic - with a choice of short or detailed, and a tone that's either warm and encouraging or direct and rigorous. The old warmth slider is gone, because nobody could tell you what the middle setting meant.
Marking is the part everyone talks about. Handing work back is the part that eats Friday afternoon.
It now starts with a plain question: 18 graded submissions. How do you want to hand it back? Then two doors.
Share it digitally. GradingPal works out how each student is reachable and shows you the plan before anything goes: nine through Google Classroom, four by email, two who can't be reached yet. Nobody gets quietly skipped. Add the missing emails right there, or print those two instead. Anything still waiting for your review gets flagged, with one tap to approve it.
Print and hand back. A PDF report per student, zipped and ready for the photocopier. No student accounts needed, which matters in a lot of elementary classrooms.
Two things here are new, and they're the ones teachers write to us about.
You can schedule the release. Pick a time and the feedback goes out then, including anything you approve between now and then. Mark on Saturday, release Monday at 8am. Teachers are using this so students read their comments in class, where someone can help them act on it, rather than alone at the weekend.
You can send feedback in the language your student reads. You set two things per assignment. The language the work is judged in - including whether that's UK or US English, so "colour" is right in one and flagged in the other. And the language the feedback comes back in: the same as the work, one language you choose for the whole class, or matched to each student automatically.
So a newcomer who writes her Gatsby essay in Spanish gets comments in Spanish that she can act on tonight, and your view stays in English, including the reasoning behind each score. You never need a translator to read your own gradebook. English (US and UK), Spanish, French, German and Brazilian Portuguese are supported, with automatic matching beyond that on a best-effort basis. Quotes from the student's own writing are never translated, because a quote is proof and proof should stay in their words.
One more small thing that changed how we think. Before you share anything, you can view as student: see the class or the assignment exactly as one student sees it, and step between students to check it reads clearly. Feedback isn't finished when it's written. It's finished when someone can read it.
A release is shaped as much by what you say no to.
We don't detect AI writing. It's the thing we get asked for most. We still won't do it. Detectors get it wrong in both directions, and the cost of being wrong lands on a student who usually can't argue back. We'd rather help you set work where the question matters less: drafts, working shown, writing done in class. Our thinking is on the responsible AI page.
We don't curve or rank. Each student is marked against your rubric, not against the class. Ask for a ranking and GradingPal will tell you no, and why.
Nothing goes out without you. There's no overnight mode, no queue that quietly sends grades to students while you sleep. We were asked for it. The moment a score reaches a student without you seeing it, it stops being your judgement, and that's the thing worth protecting.
We're not your gradebook of record. The mastery view is a teaching tool. Your district system stays the official record, and exports are there so the two can live side by side.
Trust is built from small things. We thought reassurance came from policy pages. It comes from a quote under a criterion, a "set by teacher" label, an "insufficient evidence" cell, and a suggested change you can reject. Teachers don't want to be told the AI is careful. They want to watch it be careful.
Every default is an opinion. The moment we listed five ways to work out a mastery level, it was obvious our old single method had been making a teaching decision for everyone, quietly. Naming the choices was more honest than choosing well.
Taking things away beats adding things. The change teachers mention first isn't the biggest thing we built. It's that publishing, syncing and collecting became one button. Removing an idea from someone's head is worth more than adding a feature to a list.
And one we got wrong. Our first mastery build gave a student a level as soon as there was any work at all. It demoed beautifully and it was wrong in a classroom, because a level based on one quiz starts a conversation nobody should be having yet. "Insufficient evidence (2 of 3)" is a duller screen and a better one.
Nothing you built has moved. Your classes, assignments, rubrics and graded work all carried over, and you don't need a new account. What changed is the route through them.
If you have twenty minutes this week, do these five things in order:
New to GradingPal? You don't have to start with real student work. Open a sample assignment, watch it marked, and poke at the result. The free plan gives you 50 submissions a month, any assignment type, no card.
We didn't rebuild GradingPal to make the feature list longer. We rebuilt it so it's quicker, so it handles the work you actually set - including the pile of handwritten worksheets on your desk - and so that every mastery level it shows you comes with the student's own work sitting behind it.
The full list of what changed is on the what's new page. If you already teach with us, open GradingPal and mark one assignment. If you don't, try it on a sample first and see whether you agree with what it says.
No. It's the same GradingPal underneath. Your classes, assignments, rubrics and graded work carried over unchanged, and you don't need a new account.
Essays and writing, worksheets, problem sets, quizzes, exams, lab reports and science explanations, slides, video and audio, and art coursework. Handwritten work included, scanned or photographed.
They're reviewed by our team and drawn from sources teachers already use: official state assessment writing rubrics such as MCAP, ACAP and KSA, the Smarter Balanced performance-task rubrics, AP and IB MYP rubrics, and everyday classroom rubrics across subjects and grades. Using one makes your own editable copy, and the original stays as it is for everyone else.
No. Students can work on paper. Scan the class set as one PDF, GradingPal splits it into one submission per student, and you can print the feedback and hand it back in class. No student accounts needed.
All 50 states and Washington, D.C., including Common Core and NGSS and each state's own sets, across English language arts, maths, science and social studies. Also AP, IB, Cambridge International and the UK exam boards (AQA, Edexcel, OCR and WJEC), plus WIDA and ACTFL.
No. Standards are optional. You can pick your standard sets and see plain per-standard class averages with no mastery language anywhere. Tracking levels is a separate switch.
You choose. Five methods: every piece of evidence counts equally, recent work counts more, only the last few attempts, the most frequent level, or the best level so far. Three scales, with editable cut-offs. Each result shows the calculation in a sentence, and you can override any level yourself with a reason.
Yes. Feedback can go out in the same language as the work, one language you pick for the class, or matched to each student. English (US and UK), Spanish, French, German and Brazilian Portuguese are supported, with automatic matching beyond that on a best-effort basis. Your own view stays in English, and quotes from the student's writing are never translated.
Yes. Share it now, or schedule it for a time you choose. Anything you approve before that time goes out with it.
Only if your rubric says so. Grammar findings are shown to you and your student, but scores follow your rubric.
No. Every score is a suggestion until you approve it, and nothing reaches a student until you share it.
Google Classroom is available now: import classes, rosters, co-teachers, assignments and rubrics, then post feedback back with or without grades.
Discover why human review is essential in AI grading. This guide explores the critical role of "human-in-the-loop" design for accuracy, fairness, equity, and better student outcomes in K-12 classrooms. Learn how tools like GradingPal combine powerful AI with teacher expertise.
· 13 min read
Read the article →
Discover the 6 AI-powered feedback styles in GradingPal. This complete 2026 guide shows K-12 teachers how to deliver more effective, personalized student feedback for writing and multi-media assignments with real examples and practical tips.
· 10 min read
Read the article →
Standards-based grading (SBG) focuses on mastery rather than points and averages. This definitive 2026 guide covers what standards-based grading is, its research-backed benefits, detailed alignment with Common Core and state standards (NGSS, TEKS, etc.), step-by-step implementation, rubric design, challenges, and how tools like GradingPal make it practical and effective for busy K-12 teachers.
· 14 min read
Read the article →One email when we publish something new on grading, feedback and standards. No newsletter filler.
See our privacy policy.
Reading about it only goes so far. Bring a class set you have graded and compare.
Free plan, no credit card. 50 submissions a month, any assignment type.