Skip to content
Weights Not Walls
Menu

deepseek-ai

DeepSeek-V4.1-Flash-1017

Reviewedaccepts images
Size
763B parameterscounted from the safetensors header on the Hub
Licence
unknown, MIT LicenseOur reading, check the licence: You can use, change and redistribute the model for any purpose, including commercially. Keep the copyright notice and the licence text with copies. You do not have to share your changes.
On the Hub since
17 Oct 2026the date the repo was created
Context length
1M tokensfrom the model's config
Kind
Base model (no base model declared on the Hub)DeepseekV41ForCausalLM
On Hugging Face
deepseek-ai/DeepSeek-V4.1-Flash-1017502k downloads in the last 30 days, 6.0k likes

Scores

23 independent results on 13 eval sets
Chess PuzzlesRe-run by Epoch AI
  • effort max46.0%±5.019 Oct 2026, log

Epoch's name: deepseek-v4.1-flash-1017. Best closed model on this set: claude-fable-5-2_max, 79.0% (Anthropic). All 166 results

  • effort max46.3%±7.919 Oct 2026, Epoch AI

Epoch's name: deepseek-v4.1-flash-1017. Best closed model on this set: claude-fable-5-2_max, 100.0% (Anthropic). All 83 results

  • effort max71.2%±2.719 Oct 2026, Epoch AI

Epoch's name: deepseek-v4.1-flash-1017. Best closed model on this set: claude-fable-5-2_max, 95.1% (Anthropic). All 127 results

GPQA diamondRe-run by Epoch AI
  • effort max93.2%±1.619 Oct 2026, log

Epoch's name: deepseek-v4.1-flash-1017. Best closed model on this set: gpt-6.2-astra_max, 96.8% (OpenAI). All 341 results

Mystery Game PuzzlesRe-run by Epoch AI
  • effort max49.0%±5.019 Oct 2026, Epoch AI

Epoch's name: deepseek-v4.1-flash-1017. Best closed model on this set: gpt-6.2-astra_max, 95.0% (OpenAI). All 148 results

OTIS Mock AIME 2024-2025Re-run by Epoch AI
  • effort max99.4%±0.419 Oct 2026, log

Epoch's name: deepseek-v4.1-flash-1017. Best closed model on this set: claude-fable-5_high, 100.0% (Anthropic). All 318 results

SimpleQA VerifiedRe-run by Epoch AI
  • effort max42.5%±1.619 Oct 2026, log

Epoch's name: deepseek-v4.1-flash-1017. Best closed model on this set: claude-fable-5-2_max, 94.0% (Anthropic). All 105 results

APEX-AgentsReported

Epoch's name: deepseek-v4.1-flash-1017. Best closed model on this set: gemini-4-carbon_unknown, 84.2% (Google DeepMind). All 66 results

ARC-AGIReported

Epoch's name: deepseek-v4.1-flash-1017. Best closed model on this set: claude-fable-5_max, 98.5% (Anthropic). All 318 results

ARC-AGI-2Reported

Epoch's name: deepseek-v4.1-flash-1017. Best closed model on this set: gpt-6-astra_max, 95.0% (OpenAI). All 299 results

CritPtReported

Epoch's name: deepseek-v4.1-flash-1017. Best closed model on this set: gemini-4-carbon_high, 36.0% (Google DeepMind). All 271 results

CursorBenchReported

Epoch's name: deepseek-v4.1-flash-1017. Best closed model on this set: gemini-4-carbon_high, 60.4% (Google DeepMind). All 94 results

WeirdML v3Reported

Epoch's name: deepseek-v4.1-flash-1017. Best closed model on this set: claude-fable-5-2_xhigh, 65.3% (Anthropic). All 27 results

Claimed by the maker Claimed

The model card has no claimed scores in a machine-readable form (the card's model-index), so none are shown. The maker may give scores in the card text: read the card.

Other views, which we link to and do not copy: look this model up on Artificial Analysis and the Arena leaderboard.

Independent scores from Epoch AI (CC-BY 4.0), fetched 19 Nov 2026: Epoch AI, ‘Capabilities & benchmarking’. Published online at epoch.ai. Retrieved from ‘https://epoch.ai/benchmarks’ [online resource]. Each row links its own source. What the badges mean.

Run it yourself

file sizes read from the Hub 19 Nov 2026

Memory needed = weights + context cache + 0.50 GB for the runtime. Context cache at 8k tokens: 0.63 GB, worked out from the model's config (deepseek-ai/DeepSeek-V4.1-Flash-1017), at 16-bit precision. Choose a machine above to see a verdict for each format. Method

GGUF for llama.cpp, Ollama, LM Studio and others

unsloth/DeepSeek-V4.1-Flash-1017-GGUF (converted by unsloth), 1.6M downloads. Sizes are the real files on the Hub, all parts added up. For image input you also need the projector file mmproj-F16.gguf (0.90 GB), not counted below.

FormatSizeMemory neededGet it
UD-IQ1_S217.0 GBfile size, 5 parts218.1 GB217.0 + 0.63 + 0.50
Commands
llama.cpp
llama-server -hf unsloth/DeepSeek-V4.1-Flash-1017-GGUF:UD-IQ1_S
Download
hf download unsloth/DeepSeek-V4.1-Flash-1017-GGUF --include "UD-IQ1_S/DeepSeek-V4.1-Flash-1017-UD-IQ1_S-*-of-00005.gguf" --local-dir DeepSeek-V4.1-Flash-1017-GGUF
UD-IQ1_M233.2 GBfile size, 5 parts234.3 GB233.2 + 0.63 + 0.50
Commands
llama.cpp
llama-server -hf unsloth/DeepSeek-V4.1-Flash-1017-GGUF:UD-IQ1_M
Download
hf download unsloth/DeepSeek-V4.1-Flash-1017-GGUF --include "UD-IQ1_M/DeepSeek-V4.1-Flash-1017-UD-IQ1_M-*-of-00005.gguf" --local-dir DeepSeek-V4.1-Flash-1017-GGUF
UD-IQ2_XXS279.8 GBfile size, 6 parts281.0 GB279.8 + 0.63 + 0.50
Commands
llama.cpp
llama-server -hf unsloth/DeepSeek-V4.1-Flash-1017-GGUF:UD-IQ2_XXS
Download
hf download unsloth/DeepSeek-V4.1-Flash-1017-GGUF --include "UD-IQ2_XXS/DeepSeek-V4.1-Flash-1017-UD-IQ2_XXS-*-of-00006.gguf" --local-dir DeepSeek-V4.1-Flash-1017-GGUF
UD-Q2_K_XL311.8 GBfile size, 7 parts312.9 GB311.8 + 0.63 + 0.50
Commands
llama.cpp
llama-server -hf unsloth/DeepSeek-V4.1-Flash-1017-GGUF:UD-Q2_K_XL
Download
hf download unsloth/DeepSeek-V4.1-Flash-1017-GGUF --include "UD-Q2_K_XL/DeepSeek-V4.1-Flash-1017-UD-Q2_K_XL-*-of-00007.gguf" --local-dir DeepSeek-V4.1-Flash-1017-GGUF
UD-IQ3_XXS367.6 GBfile size, 7 parts368.8 GB367.6 + 0.63 + 0.50
Commands
llama.cpp
llama-server -hf unsloth/DeepSeek-V4.1-Flash-1017-GGUF:UD-IQ3_XXS
Download
hf download unsloth/DeepSeek-V4.1-Flash-1017-GGUF --include "UD-IQ3_XXS/DeepSeek-V4.1-Flash-1017-UD-IQ3_XXS-*-of-00007.gguf" --local-dir DeepSeek-V4.1-Flash-1017-GGUF
UD-IQ3_S394.5 GBfile size, 8 parts395.7 GB394.5 + 0.63 + 0.50
Commands
llama.cpp
llama-server -hf unsloth/DeepSeek-V4.1-Flash-1017-GGUF:UD-IQ3_S
Download
hf download unsloth/DeepSeek-V4.1-Flash-1017-GGUF --include "UD-IQ3_S/DeepSeek-V4.1-Flash-1017-UD-IQ3_S-*-of-00008.gguf" --local-dir DeepSeek-V4.1-Flash-1017-GGUF
UD-Q3_K_XL451.8 GBfile size, 8 parts452.9 GB451.8 + 0.63 + 0.50
Commands
llama.cpp
llama-server -hf unsloth/DeepSeek-V4.1-Flash-1017-GGUF:UD-Q3_K_XL
Download
hf download unsloth/DeepSeek-V4.1-Flash-1017-GGUF --include "UD-Q3_K_XL/DeepSeek-V4.1-Flash-1017-UD-Q3_K_XL-*-of-00008.gguf" --local-dir DeepSeek-V4.1-Flash-1017-GGUF
UD-Q4_K_XL466.3 GBfile size, 9 parts467.4 GB466.3 + 0.63 + 0.50
Commands
llama.cpp
llama-server -hf unsloth/DeepSeek-V4.1-Flash-1017-GGUF:UD-Q4_K_XL
Download
hf download unsloth/DeepSeek-V4.1-Flash-1017-GGUF --include "UD-Q4_K_XL/DeepSeek-V4.1-Flash-1017-UD-Q4_K_XL-*-of-00009.gguf" --local-dir DeepSeek-V4.1-Flash-1017-GGUF
UD-Q8_K_XL473.5 GBfile size, 9 parts474.6 GB473.5 + 0.63 + 0.50
Commands
llama.cpp
llama-server -hf unsloth/DeepSeek-V4.1-Flash-1017-GGUF:UD-Q8_K_XL
Download
hf download unsloth/DeepSeek-V4.1-Flash-1017-GGUF --include "UD-Q8_K_XL/DeepSeek-V4.1-Flash-1017-UD-Q8_K_XL-*-of-00009.gguf" --local-dir DeepSeek-V4.1-Flash-1017-GGUF

MLX for Apple silicon

FormatSizeMemory neededRepo
MLXddalcu/DeepSeek-V4.1-Flash-1017-MLX-Serve-iQ-MLX-3.0bpw397.9 GBfile size399.1 GB397.9 + 0.63 + 0.50Download page
MLXmlx-community/DeepSeek-V4.1-Flash-1017-oQ4e-mtp402.4 GBfile size403.5 GB402.4 + 0.63 + 0.50Download page
Other conversions (14; sizes not read)

The Hub tags 100 or more quantised repos as conversions of this model; we read the most downloaded and show the maker's own first.

Run it by API

Hugging Face Inference Providers, fetched 19 Nov 2026, 13:58 UTC

35 providers serve this model through the Hugging Face router. Prices are US dollars per million tokens. Precision is what the provider reports; a lower precision than the original weights can change answers. Live first, then cheapest output.

ProviderInOutContextPrecisionTokens/sFirst tokenUptime 24hToolsJSON
ovhcloud$0$01Mnot given980.5 s?yesyes
DeepInfra$0.098$0.291Mfp81060.9 s98.5%yesyes
Pearl Research$0.05$0.301Mfp81361.1 s90.3%yesyes
StepFun$0.10$0.42977kfp81301.0 s100.0%yesyes
Ant Digital$0.12$0.481Mfp8851.6 s100.0%yesno
Novita$0.12$0.481Mfp81161.3 s100.0%yesyes
River$0.12$0.481Mnot given951.0 s100.0%yesyes
Cohere$0.15$0.501Mfp81490.8 s100.0%yesyes
Particle AI$0.06$0.501Mnot given1210.5 s100.0%yesyes
Relace$0.12$0.501Mfp41430.9 s100.0%yesyes
Wafer$0.036$0.501Mmxfp41330.4 s100.0%yesyes
Morph$0.059$0.521Mfp81150.6 s99.6%yesyes
Cloudflare$0.10$0.601Mnot given1011.9 s99.7%yesyes
DekaLLM$0.10$0.601Mnot given1431.7 s100.0%yesyes
Inceptron$0.10$0.601Mfp41390.4 s99.6%yesyes
SiliconFlow$0.15$0.601Mfp81351.1 s99.9%yesyes
Ionstream$0.21$0.621Mnot given1920.8 s99.1%yesyes
InferenceNet$0.08$0.641016kfp82680.2 s100.0%yesyes
ModelRun$0.17$0.671Mfp41120.9 s100.0%yesyes
PrimeIntellect$0.044$0.701Mnot given1120.6 s100.0%yesyes
DigitalOcean$0.18$0.721Mnot given1460.8 s100.0%yesyes
Nebius$0.18$0.721Mfp42400.5 s100.0%yesyes
Makora$0.15$0.801Mfp81541.1 s97.3%yesyes
Fireworks$0.21$0.841Mnot given1151.0 s76.6%yesyes
OpenInference$0.06$0.901Mfp41460.9 s100.0%yesyes
Mistral$0.24$0.961Mfp81541.1 s100.0%yesyes
PrimaLabs$0.24$0.961Mnot given1261.3 s99.7%yesyes
Modal$0.25$1.001Mnot given1040.9 s100.0%yesyes
BaseTen$0.30$1.201Mnot given1330.7 s100.0%yesno
CoreWeave$0.30$1.201Mfp82170.5 s100.0%yesyes
Crusoe$0.29$1.201Mfp82100.9 s98.2%yesyes
Near AI$0.30$1.201Mnot given430.7 s99.3%yesyes
Parasail$0.30$1.201Mfp82020.5 s100.0%yesyes
Together$0.30$1.201Mnot given2130.4 s100.0%yesyes
Venice$0.30$1.20977kfp81351.4 s100.0%yesyes

Through the router, with a Hugging Face token (the router bills your Hugging Face account). Change the name after the colon to pick another provider.

curl https://router.huggingface.co/v1/chat/completions \ -H "Authorization: Bearer $HF_TOKEN" -H "Content-Type: application/json" \ -d '{"model": "deepseek-ai/DeepSeek-V4.1-Flash-1017:ovhcloud", "messages": [{"role": "user", "content": "Hello"}]}'

Not built yet: one key for every open model, with BlockRun, and our own tested-quality score per provider. Until then, precision and measured speed are the best guide to a provider here.

Fine-tunes of this model

from our snapshot of the Hub

Repos that declare this model as their base, most downloaded first. Listing is not a recommendation; unreviewed ones have not been checked.

Planned for this page Not built yet

Try it in private chat
A link to Solana AI's confidential chat for the models it serves. This model is not listed as served there.
Fine-tune it
Fine-tuning a few base models from this page and publishing the result here.
Creators and earnings
Pages for people who publish fine-tunes and adapters of this model, and what they earn.
Fund and bounties
Bounties for new open models, paid from a public fund. How the fund will work.
Contribute compute
How to give compute to open model work, with Akash.

Model data from the Hugging Face Hub 19 Nov 2026, 13:43 UTC. The repo. Found a mistake on this page? Tell us.