AI Proofreader

Free.ai Proofreader is a professional-editor-style pass over your text with three depth levels: Light copy-edit (grammar / spelling / punctuation only — voice untouched), Full edit (grammar + style — weak verbs, redundancy, awkward phrasing), and Developmental (adds structural notes — buried ledes, voice consistency, thesis clarity). Pick your document type (general / academic / fiction / business / blog / technical) so the editor applies genre-appropriate conventions. Returns the edited text plus labeled notes with severity (minor / moderate / major).

Grammar Checker (/text/grammar/) is grammar-only — objective errors, fast pass. Proofreader goes further: Full mode catches style issues (weak verbs, redundancy, passive-where-active, consistency), and Developmental mode flags structural problems (buried ledes, inconsistent voice, unclear thesis). Use Grammar for quick spot-checks; use Proofreader for final-pass editing on longer work.

Grammarly Premium is $12/mo for style + clarity suggestions. ProWritingAid is $10/mo with 20+ style reports. Scribendi charges $0.019–0.089/word for human proofreading ($100+ for a 5K-word piece). Reedsy charges $400+ for a developmental edit. Free.ai is free with a daily token pool, ships all three depth levels, and runs on self-hosted GPUs. For high-stakes work (book manuscripts, legal contracts, dissertations) you still want a human editor — our Developmental mode is advisory, not a full dev edit.

Four known gaps: (1) Voice consistency across long manuscripts — AI is good at flagging voice drift but bad at matching a specific author's voice. (2) Fact-checking — the proofreader does NOT verify facts; it only flags grammar/style. Use a separate fact-checking pass. (3) Cultural sensitivity — AI can miss subtle bias or culturally-specific errors. (4) Intentional style breaks — poetic prose, experimental fiction, or dialect-heavy dialogue may get over-corrected; use Light pass for creative work.

Up to 5 structural notes per run: unclear thesis or premise, buried lede, inconsistent point-of-view, voice drift between sections, pacing issues, missing transitions, over- or under-explained sections, opening/closing weakness. Each note has a one-sentence rationale. Developmental mode is a STARTING POINT for structural review, not a complete developmental edit — human editors spend hours on each chapter for a reason.

Fiction document type explicitly tells the model to preserve dialect, voice, intentional sentence fragments, and stylistic punctuation — only flag errors, never creative choices. Still, review the diff carefully for any line where the meaning matters. Comma splices for rhythm, em-dashes for pacing, intentional sentence fragments — these should survive Fiction + Light pass.

Yes — Academic document type preserves citations verbatim (APA, MLA, Chicago, footnote markers). Hedge words and passive voice are accepted as field convention. Flag jargon only if inconsistent across the document. Always spot-check nested brackets — no AI is perfect with complex citation formats.

Seven direct picks: English, Spanish, French, German, Portuguese, Italian, Dutch. Auto-detect works across the Free.ai language set, but editorial conventions vary hugely by language so quality drops sharply for lower-resource languages. For non-listed languages use /text/grammar/ which is more forgiving.

Up to 5,000 words (~40,000 characters) per run. For book-length manuscripts, proofread chapter by chapter — quality stays highest under 8K tokens of context. Developmental notes work best on standalone pieces (single chapter, single essay, single article).

Text is sent to Free.ai's self-hosted GPUs, processed, and discarded. We do not train on user inputs. We log only timestamp + token cost for billing — never the source text. Generation history (signed-in users) stores a 200-char preview.

No — first proofreading pass runs free, no email, no credit card. Sign up for 10,000 free signup tokens + generation history + access across all 400+ Free.ai tools.

Yes. POST to /v1/chat/ with the proofreader system prompt and parse the JSON response in your client. Useful for CMS auto-edit pipelines, WordPress pre-publish checks, or batch manuscript review. Full snippets at /api/.