Sharpening the red pen…
Heads up
Document Quality
Corrected text
Suggestions
About AI grammar checking
AI grammar checking can miss context-dependent errors and sometimes flags valid sentences as problems. Always review each suggestion before applying it.
For high-stakes writing (legal, medical, academic submissions), pair this tool with human review. The audit is produced by a large language model and is intended as a fast first-pass, not a citation.
Recent audits
How it works
The AI Grammar Checker audits your text in a single pass and returns a Document Quality Score, seven sub-category sub-scores, a fully-corrected version of the text, and per-finding cards that explain every flagged issue. The audit is produced by a large language model, which can read context across long sentences and across the document as a whole, so it catches issues that rule-based checkers (LanguageTool, classic Grammarly) miss.
Input modes:
- Paste text. Most direct. Up to your plan's word cap (1k / 2k / 4k / 6k / 10k across Free to Platinum).
- URL. We fetch the page through a proxy, extract the main body, audit from there. Paywalled pages, login-gated content, and aggressive bot protection can prevent fetch; falls back to Paste mode.
- Upload. PDF or .txt file up to 10 MB. Text-based PDFs (Word, Google Docs, LibreOffice, Acrobat). Scanned image-only PDFs aren't supported.
Seven scoring sub-categories: Grammar (syntax, agreement, tense), Spelling, Punctuation, Clarity (wordiness, ambiguous reference), Style (voice, variety, flow), Tone (matches your target tone selection), and Readability (sentence and word length). Each gets its own 0-100 ring next to the overall grade.
Three-severity finding cards: High (rule violations that hurt comprehension), Medium (style improvements that strengthen the writing), Low (taste-level suggestions you can ignore). Each card shows the original passage, the suggested rewrite, and a plain-English explanation of the rule.
Honest framing. AI grammar checking can miss context-dependent errors AND sometimes flags valid sentences as problems (a "false positive"). Always review each suggestion before applying it. For high-stakes writing (legal, medical, academic submissions) pair this with human review.
Plan caps. Each call consumes a portion of your shared daily AI token budget. The budget is shared across every AI tool on this site (Free 50/day; Bronze 100; Silver 200; Gold 400; Platinum 1,000).
What you will see
- English variant (US / UK / CA / AU)
- Drives spelling and punctuation rules. "Color" vs. "Colour"; single vs. double quotes in dialog; the Oxford comma convention. Auto-detect picks one based on the text; explicit picks override.
- Target tone
- What tone you want the writing to sound like. The Tone sub-score grades how well your draft matches the target.
- Document Quality Score
- 0-100 with a letter grade (A 90+, B 75+, C 60+, D 40+, F under 40). The big circle in the score panel.
- Category rings
- Seven smaller rings next to the score, one per sub-category. Color goes blue (75+), amber (60-74), red (under 60).
- Findings count pill
- How many issues the audit flagged. Click any severity (High, Medium, Low) in the filter row to show/hide that severity in the finding stack.
- Tone label / Readability label pills
- The model's plain-English read on the actual tone of your text (which may differ from your target) and its readability level (e.g. "8th-grade reading level").
- Corrected text panel
- Your full text with every fix applied. Toggle "Show original" to compare. Copy with one click.
- Suggestions stack
- Per-finding cards with severity badge, rule name, original passage, suggested rewrite, and explanation. Use the severity filter to focus on what matters first.
- Recent audits
- The last 10 submissions, kept in your browser for 30 days.
Frequently asked questions
How do I check the grammar of my text?
Paste your text into the box (or use URL mode to fetch a webpage, or Upload mode for a PDF or .txt file), pick your English variant (US / UK / CA / AU), pick a target tone (Formal, Casual, Academic, Business), and click Check Grammar. You get a 100-point Document Quality Score with letter grade, 7 sub-category sub-scores (Grammar, Spelling, Punctuation, Clarity, Style, Tone, Readability), the fully-corrected version of your text, and per-finding cards that explain every flagged issue with severity (high / medium / low) and a plain-English rule explanation.
Is ChatGPT better than Grammarly?
They are different tools optimized for different jobs. Grammarly is a real-time editor integration with strong rule-based detection of common mistakes; ChatGPT is a general LLM that can grammar-check on demand but is not optimized for that workflow and gives no consistent severity grading or rule explanations. This tool is closer to ChatGPT under the hood (it uses a large language model to do the audit) but wraps the output in the structure Grammarly users expect: scored sub-categories, severity-graded findings, rule explanations, and a one-click fully-corrected text panel. Free, no signup.
Which grammar checker is most accurate?
No grammar checker is perfectly accurate. AI grammar checkers (this tool, ChatGPT, Grammarly Premium) are strongest on context-dependent issues like subject-verb agreement across long sentences, comma splices, and tone consistency. Rule-based checkers (LanguageTool, classic Grammarly) are strongest on mechanical errors like spelling and standard punctuation rules. Both can flag valid sentences as problems, so always review each suggestion before applying it. For high-stakes writing, pair any checker with human review.
Can AI fix grammar errors automatically?
Yes. The Corrected text panel shows your input with every flagged issue already applied. You can toggle "Show original" to see your draft, then toggle back to the corrected version. Copy the corrected text with one click. The AI also produces per-finding cards explaining what changed and why, so you can learn from the corrections rather than just accepting them blindly, useful for ESL writers and anyone who wants to improve their writing over time.
What are the most common grammar mistakes the tool catches?
- Subject-verb disagreement, especially when the subject and verb are separated by a long phrase.
- Comma splices, two independent clauses joined by a comma instead of a semicolon or period.
- Run-on sentences, ideas that should be split into multiple sentences.
- Misplaced modifiers, where the descriptive phrase isn't adjacent to what it modifies.
- Passive voice overuse, flagged when active voice would be clearer.
- Wordiness, phrases that can be shortened without losing meaning ("due to the fact that" -> "because").
- Homophone confusion, their/there/they're, your/you're, its/it's.
- Inconsistent tense, switching between past and present without reason.
Is this grammar checker really free?
Yes, no signup, no credit card, no free-trial countdown. The tool runs against a daily AI-token budget that's shared across every AI tool on the site (Free 50 tokens/day; Bronze 100; Silver 200; Gold 400; Platinum 1,000). Each audit uses 1-3 tokens. For most users the Free plan covers their daily checking comfortably. Paid plans exist for higher-volume users (content teams, ESL students doing daily work) but no feature is gated behind a paywall.
What is a Document Quality Score?
A 100-point composite that summarizes the audit across seven sub-categories: Grammar, Spelling, Punctuation, Clarity, Style, Tone, and Readability. Each sub-category has its own 0-100 score (rendered as a ring next to the overall grade). The overall score is the model's holistic assessment, not a mechanical roll-up of the sub-scores, so it can occasionally rate a piece higher than its weakest sub-score (when the overall communication still works) or lower (when the sub-scores look fine but the piece doesn't hang together).
How long can the input be?
Word caps per submission: Free 1,000 words; Bronze 2,000; Silver 4,000; Gold 6,000; Platinum 10,000. We chose those numbers because the audit duplicates the input in its corrected_text output, so the binding constraint is the model's output token ceiling, not its input ceiling. For text over your plan's cap, the input is truncated at the word boundary with a banner showing how much was dropped. Split very long documents into chapters or sections and audit each separately.