DeepSeek V4.1 Flash ranks #17 of 87 models on the November 2026 ToneBench board through its best configuration, DeepSeek V4.1 Flash (max). This setting, DeepSeek V4.1 Flash (no thinking), is #82 of 151 configurations, with a Writing Score of 76.8/100 and a Writing Elo of 1562. One script costs $0.0037 at API list price.
Last updated , from board data generated at 17:52 UTC.
Rank, scores, the nine metrics, the 10 reference scripts, cost and setup for one
configuration (a model at one reasoning setting). Key Results sums it up; the section links below it jump to the detail.
Model Rank ?
#17 of 87 modelsthrough DeepSeek V4.1 Flash (max); this setting is #82 of 151 configurations
Writing Score ?
76.8 out of 100
Writing Elo ?
1562 95% range 1514 to 1605
Cost per Script
$0.0037 at API list price
Key Results
StandingDeepSeek V4.1 Flash ranks #17 of 87 models through its best configuration, DeepSeek V4.1 Flash (max). This setting is #82 of 151 configurations by Writing Elo (1562), in the middle third of the board. Its 95% range (1514 to 1605) overlaps 20 other configurations, ranked #71 to #91.
MetricsBest-ranked: Tone & Voice Match (76.0, 62nd of 151 configurations, +1.9 against the board median). Lowest-ranked: Visual Cue Quality (65.8, 115th of 151 configurations, -9.0 against the board median).
ScriptsFrom 70.2 on Script 10 (engineering process walkthrough) to 80.5 on Script 7 (personal technical walkthrough): a 10.3-point spread across the 10 reference scripts.
ConsistencyIts Writing Score moves ± 3.3 between drafts of the same script, against ± 2.5 for the median configuration: typical.
ValueGPT-6 Luna (medium) ranks higher with an overlapping Elo range and costs about $0.0026 per script, against $0.0037 per script for this one.
Writing Elo comes from comparing its average on each script with every other configuration's average on that same script; a gap smaller than the draft-to-draft variation counts as a draw (what the ranges can and cannot show).
Visual Cue Quality ranks in the bottom half (115th of 151 configurations), so expect to add most of the visual cues by hand.
The script format moves its score a lot. If most of your scripts look like its best or weakest format, that result is a better guide than the average.
It outranks 56 configurations that cost at least 50% more.
Among open-weight models, DeepSeek V4.1 Flash is 5th of 34. This setting is 13th of 43 open-weight configurations. The best open-weight configuration is GLM-5.3 (max), at configuration #22, and every configuration above it is proprietary.
Lab
DeepSeek 2nd of 5 modelsthis setting 8th of 11 configurations
Weights
Open 5th of 34 modelsthis setting 13th of 43 configurations
Consistency
± 3.3 typical spread
Average Tokens
16.3k input + output
Time per Script
0.3 min average, retries included
Highest-Ranked in This LabDeepSeek V4.1 Flash 1017 (max) ranks highest of the DeepSeek configurations by Writing Elo, with a Writing Score +4.8 against this configuration, for about 1 cent more per script.
Each corner is one metric on a 0 to 100 scale; the further out, the higher the score. A corner outside the dashed line beats the board median. Hover or tap a point for its exact score, and a metric name for what it rewards.
DeepSeek V4.1 Flash (no thinking)Board medianBoard best per metricMax possible (100)
Scores by Metric
Nine metrics, each scored 0 to 100 and weighted into the Writing Score; the weight sits under each name. Rank is out of all 151 ranked configurations.
Score colors:Strong: 82 and upMiddle: 65 to 81.9Weak: under 65
This modelBoard best (upper line)Board median (lower line)±: spread between drafts?: what the metric rewards
Tone & Voice Match ?19% weight · +1.9 vs median
board best 82.8 · Claude Opus 5.5 (max effort)76.0± 2.0 · 62ndboard median 74.0
Writing Craft & Clarity ?13% weight · +0.7 vs median
board best 83.2 · Claude Opus 5.5 (xhigh)75.7± 1.7 · 67thboard median 75.0
Substance, Accuracy & Value ?15% weight · -3.4 vs median
board best 88.4 · Claude Opus 5.5 (xhigh)78.4± 2.4 · 99thboard median 81.8
Continuity & Emotion ?14% weight · +0.2 vs median
board best 85.7 · Claude Opus 5.5 (max effort)77.8± 1.6 · 72ndboard median 77.5
YouTube Best Practices ?12% weight · -2.7 vs median
board best 84.5 · Claude Fable 5.2 (max effort)73.6± 3.1 · 90thboard median 76.4
Hook Strength ?10% weight · -0.6 vs median
board best 84.0 · Claude Fable 5.2 (xhigh)79.0± 3.1 · 87thboard median 79.6
Length Adherence ?8% weight · -4.0 vs median
board best 99.0 · Claude Fable 5.2 (xhigh)80.8± 10.5 · 88thboard median 84.8
Slop Score ?5% weight · -0.5 vs median
board best 89.0 · GPT-6 Sol (medium)80.8± 2.0 · 81stboard median 81.3
Visual Cue Quality ?4% weight · -9.0 vs median
board best 85.9 · Claude Opus 5.5 (max effort)65.8± 7.0 · 115thboard median 74.8
Scores by Reference Script
Every ranked configuration writes five drafts for each of the same 10 briefs, and each score is the mean of those drafts. The rank and board median compare it with every configuration on that script, which separates a weak result from a hard script. The result on the format closest to your own work is the most relevant.
Script 1
opinion / warning explainer
76.0
out of 100
86th of 151 configurations on this script · board median 77.2
Script 2
news-analysis / skeptical explainer
79.1
out of 100
68th of 151 configurations on this script · board median 78.4
Script 3
personal roadmap / opinion
75.7
out of 100
91st of 151 configurations on this script · board median 77.2
Script 4
short explainer
74.5
out of 100
83rd of 151 configurations on this script · board median 75.7
Script 5
news-analysis / opinion explainer
79.0
out of 100
66th of 151 configurations on this script · board median 78.0
Script 6
founder announcement / personal origin story
76.8
out of 100
97th of 151 configurations on this script · board median 79.2
Script 7
personal technical walkthrough / agentic workflow case study
80.5
out of 100
46th of 151 configurations on this script · board median 77.9
Script 8
product release announcement / personal observation
77.9
out of 100
86th of 151 configurations on this script · board median 79.0
Script 9
career guide / hiring analysis
78.5
out of 100
83rd of 151 configurations on this script · board median 79.3
Script 10
engineering process walkthrough / presentation adaptation
70.2
out of 100
99th of 151 configurations on this script · board median 75.0
Consistency Between Drafts
Every score on this page is the mean of 5 drafts per script. The ± is the standard deviation between those drafts: the larger it is, the more two drafts of the same script can differ.
Compared with the typical model, it varies most on Visual Cue Quality: ± 7.0 between drafts against ± 2.9, so two drafts of the same script can score quite differently there.
Consistency by Metric
Metric
This model
Board median
Verdict
Tone & Voice Match ?
76.0 ± 2.0
± 1.9
typical
Writing Craft & Clarity ?
75.7 ± 1.7
± 1.5
typical
Substance, Accuracy & Value ?
78.4 ± 2.4
± 2.1
typical
Continuity & Emotion ?
77.8 ± 1.6
± 1.5
typical
YouTube Best Practices ?
73.6 ± 3.1
± 2.4
typical
Hook Strength ?
79.0 ± 3.1
± 2.7
typical
Length Adherence ?
80.8 ± 10.5
± 6.6
more variable than most
Slop Score ?
80.8 ± 2.0
± 1.6
typical
Visual Cue Quality ?
65.8 ± 7.0
± 2.9
more variable than most
Cost, Time and Tokens per Script
Logged on every run; none of these figures affects the Writing Score. Time is the average time to one accepted script, retries included. Cost is one script at the API list price.
Time per Script
0.3 minfaster than the board median of 1.5 min
Prompt tokens in
13.4kstyle guide + brief + research packet
Tokens Out
2.8kscript + any reasoning tokens
Cost per Script
$0.0037measured from actual billed tokens; varies ± 0.001 between drafts
List Price Used
$0.15 / $0.6input / output, per million tokens
Scores by Judge
Four judges score the same 50 stored drafts blind against the same rubric: three LLM judges from three labs, plus Jev, a typed judge from TypeSafe that rates each metric on five described quality levels. No single judge's taste decides the result. The Writing Score of 76.8 combines their scores with the parts computed in code (Length Adherence and half of the Slop Score). The highest and lowest judge are 13.9 points apart on its Writing Score. That is close to the board median of 13.4. Each judge also has its own scale, so every card shows that judge's score against its average across the board. How the panel works.
Claude Opus 5.5 (medium)
Anthropic
71.6
+3.0 against its board average of 68.7
GLM 5.3 Flash
Z.ai
85.6
+2.7 against its board average of 82.9
GPT-6.1 Sol (medium)
OpenAI
71.7
-0.2 against its board average of 71.9
Jev 1.13
TypeSafe
78.3
+4.3 against its board average of 74.0
Evaluation Setup
How this exact configuration was run and judged: route, model ID, dates, sampling and prompt.
Exact model / route iddeepseek/deepseek-v4.1-flash at reasoning level: none.
Generations5 independent runs × 10 scripts = 50 drafts, stored after generation and never edited.
Run dates2026-10-08.
JudgesClaude Opus 5.5 at medium effort through the Claude Code subscription, GPT-6.1 Sol at medium reasoning through the Codex CLI, GLM 5.3 Flash through OpenRouter, and Jev 1.13 from TypeSafe. One call per judge per draft. All four score every draft on their own, blind to which model wrote it. The three LLM judges use the identical rubric and have to quote evidence before they score a metric. Jev answers the same rubric as one question per metric, choosing among five described quality levels, and writes no text. Seven metrics are the plain average of the four judges. Length is computed from the spoken word count, and Slop is half computed (lexical scans minus the spoken-dash penalty) and half judged. The Writing Elo is cross-checked with a Bradley-Terry fit of the same pairwise results (methodology).
Sampling and promptThe provider's default sampling, with no fine-tuning and no system-prompt changes. Every current-ranked model gets the same prompt, about 13,000 tokens (the count shifts a bit with each model's tokenizer): a style guide built from Louis's voice and storytelling guides, the target length, and the brief with its research packet. Generation is closed-book on every route: the prompt is the only source, with no web search, browsing, code execution, file access or other tool (generation).
What stays privateThe exact rubric and any non-public reference material. Each script is normally benchmarked before its video is published, which lowers the odds that a model has seen the reference, though it does not rule it out (why).
Frequently Asked Questions
How Good Is DeepSeek V4.1 Flash (no thinking) at Writing?
DeepSeek V4.1 Flash ranks #17 of 87 models on ToneBench through its best configuration, DeepSeek V4.1 Flash (max). This configuration is #82 of 151 configurations, in the middle third of the board, with a Writing Elo of 1562 and a Writing Score of 76.8/100. The ranks come from full YouTube scripts written in the voice of Louis-François Bouchard and judged blind, with the channel's finished script as the reference. The benchmark measures long-form scripted writing, so the result does not carry over to every kind of text.
Is DeepSeek V4.1 Flash (no thinking) the Best DeepSeek Model for Writing?
No. DeepSeek V4.1 Flash is 2nd of 5 DeepSeek models. The top one is DeepSeek V4.1 Flash 1017, through DeepSeek V4.1 Flash 1017 (max), about 463 Elo ahead of this setting.
Is DeepSeek V4.1 Flash (no thinking) Good Value for the Money?
It costs about $0.0037 per script. GPT-6 Luna (medium) ranks higher with an overlapping Elo range and costs less.
What Are the Strengths and Weaknesses of DeepSeek V4.1 Flash (no thinking)?
Its best-ranked metric is Tone & Voice Match (76.0, 62nd of 151 configurations, +1.9 against the board median), and its lowest-ranked is Visual Cue Quality (65.8, 115th of 151, -9.0). The Scores by Metric section shows the board median and the board's best for each one.
How Was DeepSeek V4.1 Flash (no thinking) Evaluated?
It was run through OpenRouter, with the exact model/route ID deepseek/deepseek-v4.1-flash at reasoning effort none, first run on 2026-10-08. It wrote each script 5 times with the provider's default sampling and no fine-tuning. The Evaluation Setup section and the methodology page have the details.
Does DeepSeek V4.1 Flash (no thinking) Have Open Weights?
Yes. Its weights are public, so it can be downloaded and run on your own hardware. Check the license if you need open source in the strict sense. Among open-weight configurations it is 13th of 43 for writing.