ds DeepSeek V4.1 Flash (max)
DeepSeek · open weights · writing benchmark
The short version
DeepSeek V4.1 Flash (max) sits at #22 of 148 on ToneBench, in the middle of the pack, with a writing Elo of 2085 and an overall score of 86.8 out of 100. We measured it by having it write all 10 of our real YouTube scripts, five times each, then scoring every draft blind against our own finished versions. Here is how it shook out.
Inside DeepSeek, this is the strongest writer we tested. It edges out the other 11 DeepSeek configs on the board, so if you're staying in this family for voice work, this is the one to reach for.
Among open-weights models, it comes in 2nd of 52. That is a genuinely strong showing for an open-weights model on a voice-and-tone task, which is usually where the closed frontier still pulls ahead.
Worth knowing before you rely on it: this model is swingy. Its overall score moves ± 6.9 points between runs of the same brief, versus a board median of ± 3.1. 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 9th on that dimension at 90.0, well above the board average. Its softer spot is Visual Cues (82.3, 52nd), which is the thing to watch if that metric matters most for your use.
It was uneven across the 10 scripts. Its best run was the product release announcement / personal observation (88.9) and its weakest was the engineering process walkthrough / presentation adaptation (75.2), a spread of about 14 points. Worth knowing if your writing skews toward one of those formats.
On cost, it runs about $0.012/task. For that money it beats 78 models that cost noticeably more.
Skill profile
The shape is the story: the further a corner reaches, the stronger that dimension. Biggest edge over the board: Length Discipline (+21.1); smallest edge: Substance & Value (+5.9).
Per-metric scores
Nine writing dimensions, each scored 0–100 and blended by the editorial weight shown. Rank is against all 148 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 10 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, DeepSeek V4.1 Flash (max) varies by ± 6.9 points between runs versus a board median of ± 3.1, so it is swingier than the typical model here, worth knowing if you need repeatable output.
Its most volatile dimension is Substance & Value (± 3.7 vs a board median of ± 2.3): two runs of the same brief can land visibly different substance & value scores.
| Metric | This model | Board median | Verdict |
|---|---|---|---|
| Tone & Voice ? | 87.4 ± 2.4 | ± 2.0 | typical |
| Writing Craft ? | 87.3 ± 2.6 | ± 1.9 | typical |
| Substance & Value ? | 85.2 ± 3.7 | ± 2.3 | swingier than most |
| Flow & Emotion ? | 84.9 ± 3.7 | ± 2.3 | swingier than most |
| YouTube Structure ? | 84.9 ± 4.4 | ± 3.4 | typical |
| Hook ? | 88.9 ± 1.9 | ± 2.4 | typical |
| Length Discipline ? | 90.0 ± 7.1 | ± 9.4 | typical |
| Anti-Slop ? | 91.0 ± 2.3 | ± 2.2 | typical |
| Visual Cues ? | 82.3 ± 5.0 | ± 4.2 | typical |
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 86.8 is the consensus of three family-disjoint judges scoring the same 50 stored drafts blind with the identical rubric. The highest and lowest judge differ by 4.2 points on its overall. The three judges essentially agree on this model. How the panel works: methodology.
How we ran DeepSeek V4.1 Flash (max)
- Accessed viaOpenRouter · model page ↗
- Exact model / route id
deepseek/deepseek-v4.1-flashatreasoning_effort: max - Generations5 independent runs × 10 scripts = 50 drafts, stored immutably after generation
- Run dates2026-09-10
- 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 DeepSeek V4.1 Flash (max) at writing?
On ToneBench it ranks #22 of 148 with a writing Elo of 2085 and an overall score of 86.8/100. That score comes from writing our 10 real YouTube scripts five times each and scoring every draft blind against our own finished versions across nine writing dimensions.
Is DeepSeek V4.1 Flash (max) the best DeepSeek model for writing?
Yes. Among the DeepSeek configs we tested, DeepSeek V4.1 Flash (max) is the strongest writer on the board.
Is DeepSeek V4.1 Flash (max) good value for the money?
It costs about $0.012/task. Nothing meaningfully cheaper outscores it, which puts it on the value side of the board.
What are DeepSeek V4.1 Flash (max)'s strengths and weaknesses?
Its strongest dimension is Length Discipline (9th on the board, 90.0). Its weakest is Visual Cues (52nd on the board, 82.3). The full nine-metric breakdown is on this page.
Does DeepSeek V4.1 Flash (max) write some formats better than others?
Yes. Its best of our 10 scripts was the product release announcement / personal observation (88.9) and its weakest was the engineering process walkthrough / presentation adaptation (75.2).
How consistent is DeepSeek V4.1 Flash (max) between runs?
We run every script 5 times. DeepSeek V4.1 Flash (max)'s overall score varies by about ±6.9 points between runs, versus a board median of ±3.1. That is swingier than typical, so expect more draft-to-draft variation. The full per-metric variability table is on this page.
How was DeepSeek V4.1 Flash (max) evaluated?
Via OpenRouter using the exact model/route id deepseek/deepseek-v4.1-flash at reasoning_effort max, run on 2026-09-10. 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 DeepSeek V4.1 Flash (max) open source?
Yes, it is an open-weights model. Among open-weights models it ranks 2nd of 52 for writing.