GPT-4o mini
OpenAI · proprietary · writing benchmark
The short version
GPT-4o mini sits at #121 of 130 on ToneBench, toward the value end of the board, with a writing Elo of 384 and an overall score of 55.8 out of 100. We measured it by having it write all 9 of our real YouTube scripts, five times each, then scoring every draft blind against our own finished versions. Here is how it shook out.
Within OpenAI it ranks 36th of 36. GPT-5.6 Sol (ultra) is the family's top writer here, about 1932 Elo ahead, so this config trades some quality for whatever it saves you in cost or speed.
Among proprietary models, it comes in 83rd of 83. That ranking is against the stronger half of the board: closed models still set the pace on voice fidelity here.
Worth knowing before you rely on it: this model is swingy. Its overall score moves ± 6.3 points between runs of the same brief, versus a board median of ± 3.8. A great draft and a mediocre one can come from the identical prompt. The shaded bands on the metric bars below show where that volatility lives.
What it does best is Length Discipline: it ranks 106th on that dimension at 47.9, well below the board average. Its softer spot is Anti-Slop (60.4, 128th), which is the thing to watch if that metric matters most for your use.
It was uneven across the 9 scripts. Its best run was the founder announcement / personal origin story (65.8) and its weakest was the personal technical walkthrough / agentic workflow case study (46.8), a spread of about 19 points. Worth knowing if your writing skews toward one of those formats.
On cost, it runs about $0.002/task. Cohere Command A Reasoning scores higher for less money, at $0/task, so on pure value this config is not the frontier.
Skill profile
The shape is the story: the further a corner reaches, the stronger that dimension. Biggest edge over the board: Substance & Value (-14.2); furthest behind: Visual Cues (-47.3).
Per-metric scores
Nine writing dimensions, each scored 0–100 and blended by the editorial weight shown. Rank is against all 130 current-ranked models. The thick bar is this model; the thin lines above and below are the current field's best model (darker) and average (lighter) on the same scale. The small ± number is its run-to-run variation.
Per-article scores
The same 9 real scripts every current-ranked model writes, scored individually. Different formats stress different skills.
Consistency: run-to-run variability
Every cell on this page is the mean of 5 independent runs per script; the ± numbers are the run-to-run standard deviation. Overall, GPT-4o mini varies by ± 6.3 points between runs versus a board median of ± 3.8, so it is swingier than the typical model here, worth knowing if you need repeatable output.
Its most volatile dimension is Tone & Voice (± 3.4 vs a board median of ± 2.2): two runs of the same brief can land visibly different tone & voice scores.
| Metric | This model | Board median | Verdict |
|---|---|---|---|
| Tone & Voice ? | 57.5 ± 3.4 | ± 2.2 | swingier than most |
| Writing Craft ? | 62.0 ± 3.2 | ± 2.0 | swingier than most |
| Substance & Value ? | 64.5 ± 4.2 | ± 2.6 | swingier than most |
| Flow & Emotion ? | 53.7 ± 4.4 | ± 2.6 | swingier than most |
| YouTube Structure ? | 44.7 ± 5.6 | ± 3.6 | swingier than most |
| Hook ? | 62.9 ± 5.2 | ± 3.0 | swingier than most |
| Length Discipline ? | 47.9 ± 6.6 | ± 8.6 | typical |
| Anti-Slop ? | 60.4 ± 5.8 | ± 2.6 | swingier than most |
| Visual Cues ? | 27.2 ± 9.8 | ± 4.7 | swingier than most |
Measured, not modeled
Numbers we log on every run and rarely talk about. None of these affect the writing scores; cost and latency are informational.
What each judge scored it
The published overall of 55.8 is the consensus of three family-disjoint judges scoring the same 45 stored drafts blind with the identical rubric. The highest and lowest judge differ by 9.0 points on its overall. A typical amount of judge disagreement for this board. How the panel works: methodology.
How we ran GPT-4o mini
- Accessed viaOpenAI API · model page ↗
- Exact model / route id
gpt-4o-mini-2024-07-18(provider default reasoning) - Audited route overrideOpenRouter ·
openai/gpt-4o-mini-2024-07-18forautomated_video_workflow(OpenRouter gateway for exact legacy OpenAI model; Task-7 exact OpenRouter endpoint pin openai)
OpenRouter ·openai/gpt-4o-mini-2024-07-18forai_engineer_hiring_2026, mentorship_release({'original_provider': 'openai', 'original_model': 'gpt-4o-mini-2024-07-18', 'provider': 'openrouter', 'model': 'openai/gpt-4o-mini-2024-07-18'}) - Generations5 independent runs × 9 scripts = 45 drafts, stored immutably after generation
- Run dates2026-07-29 → 2026-08-11
- Judges3-judge panel: Claude Opus 5, high effort (Anthropic) + DeepSeek V4 Flash (OpenRouter) + GPT-5.6 Sol, medium reasoning (Codex CLI): every draft scored independently by all three with the identical rubric, blind to model identity, evidence quotes required before each metric score; published score = per-metric consensus mean, verified with a Bradley-Terry refit (methodology)
- Samplingprovider-default sampling, no fine-tuning, no system-prompt changes; every current-ranked model gets the identical brief, style contract and research packet
- What stays privatethe exact rubric and any non-public reference material; source-publication timing is disclosed for exceptions (why)
Frequently asked questions
How good is GPT-4o mini at writing?
On ToneBench it ranks #121 of 130 with a writing Elo of 384 and an overall score of 55.8/100. That score comes from writing our 9 real YouTube scripts five times each and scoring every draft blind against our own finished versions across nine writing dimensions.
Is GPT-4o mini the best OpenAI model for writing?
Not quite. Within OpenAI it ranks 36th of 36; GPT-5.6 Sol (ultra) is the family's best writer here.
Is GPT-4o mini good value for the money?
It costs about $0.002/task. Cohere Command A Reasoning scores higher for less, so it is not the value pick.
What are GPT-4o mini's strengths and weaknesses?
Its strongest dimension is Length Discipline (106th on the board, 47.9). Its weakest is Anti-Slop (128th on the board, 60.4). The full nine-metric breakdown is on this page.
Does GPT-4o mini write some formats better than others?
Yes. Its best of our 9 scripts was the founder announcement / personal origin story (65.8) and its weakest was the personal technical walkthrough / agentic workflow case study (46.8).
How consistent is GPT-4o mini between runs?
We run every script 5 times. GPT-4o mini's overall score varies by about ±6.3 points between runs, versus a board median of ±3.8. That is swingier than typical, so expect more draft-to-draft variation. The full per-metric variability table is on this page.
How was GPT-4o mini evaluated?
Via OpenAI API using the exact model/route id gpt-4o-mini-2024-07-18, run on 2026-07-29. 5 runs per script, provider-default sampling, no fine-tuning; every draft scored blind with a fixed rubric by a three-family judge panel. Full details in the 'How we ran it' section and the methodology.
Is GPT-4o mini open source?
No, it is a proprietary (closed-weights) model. Among proprietary models it ranks 83rd of 83 for writing.