whatmodel.sh / DeepSeek V4.1 Flash 1017
DeepSeek V4.1 Flash 1017
↗ share
+ contribute
DeepSeek V4.1 Flash 1017 is an open-weights language model from DeepSeek, released October 2026. Licence: mit.
Model information sources
Score within the LMArena (text) comparison set only.
$ how it compares
The LLMs ranked either side of DeepSeek V4.1 Flash 1017, out of 882 with a score, then more from DeepSeek.
| # | model | score | where | ||
|---|---|---|---|---|---|
| 13 |  Claude Mythos Preview (limited preview) | 80±1 | limited preview | ||
| 14 |  GPT-5.4 Pro | 79±1 | api $67.50/M | ||
| 15 | DeepSeek V4.1 Flash 1017 | 79±1 | api $0.26/M | ||
| 16 |  Muse Wonder | 79±1 | api $8.00/M | ||
| 17 | GPT-5.5 | 78±0 | api $11.25/M | ||
| more from DeepSeek | |||||
| 24 | DeepSeek V4.1 Flash | 76±1 | local | ||
| 35 | DeepSeek V4 Pro (0813) | 75±1 | api $1.63/M +local | ||
| 39 | DeepSeek V3.2 Speciale | 74±3 | comparison only | ||
$ what it's good at
agent tasks▁▁▃▆█▅▅▅▅▄▃▁ #11 of 310 · 39% of its score · 4 tests
Beats Muse Wonder Full, GPT-5.5 Pro, GPT-5.6 Sol here, though they rank higher overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Terminal-Bench 4.0 lab | 42.3% ✎ | ▁·▁··▁▆▆▅█▇▂ beats 22/31 | 19% |
| Chartography with tools lab | 83.0% ✎ | ▂········▄█▆ beats 3/9 | 13% |
| DeepSWE v1.1 (resolved) lab | 77.0% ✎ | ▂····▂▁▁▆███ beats 30/31 | 4.3% |
| Agents' Last Exam (metric/harness unspecified) lab | 35.4% ✎ | ▂·▂·█▄▃··▂▄▄ beats 8/13 | 3.9% |
coding▁▁▁▃▆▇█▇▅▄▃▁ #13 of 466 * · 40% of its score · 2 tests
Beats GPT-5.6 Sol, Claude Fable 5 here, though they rank higher overall. Trails Qwen3.8 Max (0902), Qwen3.8 Max (2.4T-A95B) here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Arena WebDev (Code Arena) | 1664.3 ✎ | ▁▁▃▃▆█▄▇▅▅▁▁ beats 115/126 | 36% |
| DeepSWE v1.1 (resolved) lab | 77.0% ✎ | ▂····▂▁▁▆███ beats 30/31 | 4.3% |
other tests4 not in a task · 24% of its score
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Humanity's Last Exam lab | 42.6% ✎ | ▁▁▁▄▅▆▆██▆▁▃ beats 95/108 | 7.8% |
| AutomationBench v1.0.6 lab | 60.5% ✎ | ▃···▃·▃▃··█▅ beats 8/8 | 6.3% |
| NL2Repo lab | 68.2% ✎ | ▂▂▂·▃▃▅█▃▂▅▃ beats 22/22 | 5.7% |
| BabyVision with tools lab | 92.5% ✎ | █······█··██ beats 3/3 | 4.4% |
$ where to run it
Through an API
api · $0.26/M · $0.15 in / $0.6 out per 1M tokens · 1M context · api-docs.deepseek.com ↗ · price sourceUSD per 1M tokens, off-peak uncached input/output. Peak rates are $0.30/$1.20, 01:00–04:00 and 06:00–10:00 UTC weekdays excluding Chinese public holidays. Cached input excluded. API deepseek-flash serves the 1017 checkpoint; old Flash aliases also route here. Observed 2026-10-21.
Weights and documentation: huggingface.co/deepseek-ai ↗
$ how the score was worked out · DeepSeek V4.1 Flash 1017 mode… 64%, LMArena leaderboard dataset 36%
Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works
| source | results | share of the score |
|---|---|---|
 DeepSeek V4.1 Flash 1017 model card · lab-reported | 8 | 64% |
 LMArena leaderboard dataset | 1 | 36% |
| prior · pull toward the middle | 1% | |
| weighted average | = 79.1 |
$ improve this page
Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.
data updated · sources and licences · json