Thinking Levels · DeepSeek
DeepSeek V4.1 Flash 1017: Writing Quality by Reasoning Effort
Last updated , from board data generated at 16:59 UTC.
This page compares DeepSeek V4.1 Flash 1017 at 2 reasoning-effort settings, each tested on the same 10 scripts: the Writing Score (the 0 to 100 consensus of the judges and code), the Writing Elo (a rating from pairwise comparisons of those scores), and the cost and time of each setting. As a model, DeepSeek V4.1 Flash 1017 ranks #13 of 81 models through its best setting. The other ranks on this page are configuration ranks: one setting's place among all 134 configurations.
What More Effort Changes for DeepSeek V4.1 Flash 1017
Max effort gives DeepSeek V4.1 Flash 1017 a small lead over high, and the board can't confirm it. The model ran at two levels: high, which is what DeepSeek uses when the request sets no level, and max, the deepest level on offer. Max comes out 37 Elo ahead, enough to win about 55% of its script comparisons against high, but the two ranges overlap. The extra thinking stays cheap, though. Max thinks a lot longer, yet a script costs $0.014 and comes back in about 1.3 minutes, against $0.0095 and 0.9 minutes at high.
Recommended Setting
I'd run it at max. It's #24 on the whole board, and most of its edge over high is length: the scripts land closer to their target, and the visual cues tick up a little. Tone, craft, substance and hook barely move, so you get the same voice at both levels. On a bill this small, paying 1.5 times as much for that is an easy call for me.
If you're generating at volume or want scripts back faster, high is a fine pick too, at #27 and $0.0095 a script. Whichever level you choose, use this snapshot over the earlier one. At high, it lands 97 Elo above DeepSeek V4.1 Flash (high), with ranges that don't overlap. At max, the board can't separate it from DeepSeek V4.1 Flash (max), but it costs less and comes back faster.
Every Setting on the Chart
Each point is one effort setting of DeepSeek V4.1 Flash 1017, joined from the lowest setting to the highest. Higher and further left is better: a higher score for less money or less time.
How to Use the Chart
- A hollow marker is the provider default: the level the provider applies when the request sets no effort level.
- The tabs choose the score on the vertical axis, X axis switches between cost and time, and Table lists the same numbers as rows.
- Select a point for its numbers and a link to its scorecard.
On a small screen, Table lists the same numbers as rows.
All Effort Settings, Step by Step
| Setting | Writing Elo | Step | Writing Score | Cost / Script | Time / Script | Reasoning Tokens | Configuration Rank |
|---|---|---|---|---|---|---|---|
| high | 2050 ±46 | - | 81.1 | $0.0095 | 0.9 min | 9,728 | #27 |
| max | 2087 ±50 | +37 | 81.6 | $0.014 | 1.3 min | 16,960 | #24 |
The 95% Writing Elo ranges overlap, so the board cannot separate these two: DeepSeek V4.1 Flash 1017 (high) runs from 2001 to 2092, and DeepSeek V4.1 Flash 1017 (max) from 2039 to 2140.
Other DeepSeek Models by Effort
Writing Elo comes from comparing every pair of models on every script, which puts these settings on the same scale as the leaderboard. Step is the Writing Elo change from the previous explicit setting. Cost is one full script at API list prices, and time is the average for one accepted script, retries included. The thinking-levels page covers the other models, and the DeepSeek family page lists every DeepSeek model side by side.