Checked against the OpenRouter public models feed. Prices are copied, not estimated.

The Best Model ForWhich AI model to use, from ten labs.

DeepSeek

Short answer

DeepSeek's rows on this feed include public weights ids. The October 2026 V4.1 Flash snapshot is the reviewed default: it accepts images and is listed cheaper than the April V4 Pro row. The name Pro does not, on this feed, mean a higher third-party index.

Lab site · Filter the full catalogue to DeepSeek

What the top models are good for

1. DeepSeek V4.1 Flash 1017

Default · $0.15 in · $0.60 out · 1,048,576 tokens

Guide note · 19 Nov 2026

This is the 17 October 2026 snapshot of DeepSeek V4.1 Flash. It is open weights, accepts images, lists about a million tokens of context, and is priced like a budget model. On this feed its third-party intelligence index is higher than the April V4 Pro row. That index is not our measurement. It is a reason to try this snapshot first.

  • Low rate and image input
  • Open weights
  • Long context

What it is good for

Default DeepSeek for a coding task or a photo

Start here before you reach for a model with Pro in the name. The October snapshot sees images and costs less than the April Pro row. Keep a test set so you notice if a later snapshot regresses.

DeepSeek V4.1 Flash 1017 is listed at $0.15 input and $0.60 output per million tokens, with a 1,048,576-token context window.

Summarise a long document you are allowed to send out

The window is about a million tokens. Open weights are there if you later need to host it. The summary should quote the document.

DeepSeek V4.1 Flash 1017 is listed at $0.15 input and $0.60 output per million tokens, with a 1,048,576-token context window.

The tasks are editorial illustrations, not measured benchmarks. The prices in them are read from today's feed.

2. DeepSeek V4 Pro 0423

Large · $0.90 in · $1.50 out · 1,048,576 tokens

Guide note · 19 Nov 2026

OpenRouter describes the April V4 Pro row as a very large mixture-of-experts for advanced reasoning and coding, with open weights and a million-token window. Do not read the name as a leaderboard. On this feed its third-party intelligence index is lower than the October V4.1 Flash snapshot, and the token rate is higher.

  • Large open-weight reasoning model
  • Long context
  • A model to compare against Flash, not to assume is better

What it is good for

A second pass when Flash was shallow

Run your real prompt on V4.1 Flash first. Move the failures to V4 Pro. That is the honest use of a model whose name overclaims what this feed shows.

DeepSeek V4 Pro 0423 is listed at $0.90 input and $1.50 output per million tokens, with a 1,048,576-token context window.

A long technical memo, text only

The window and the reasoning switch in the feed fit a memo you paste in. There is no image input on this row.

DeepSeek V4 Pro 0423 is listed at $0.90 input and $1.50 output per million tokens, with a 1,048,576-token context window.

The tasks are editorial illustrations, not measured benchmarks. The prices in them are read from today's feed.

3. DeepSeek V4 Flash 0423

Budget · $0.09 in · $0.18 out · 1,048,576 tokens

Guide note · 19 Nov 2026

The April 2026 V4 Flash row is the cheapest reviewed text model with a million-token window and open weights. It does not accept images. Use it for bulk text.

  • Very low text rate
  • Open weights
  • Long context

What it is good for

Cheap bulk customer-email triage

Text in, a small label set, a huge window if a thread is long. Sample the errors. This is the budget end of the reviewed set for plain text.

DeepSeek V4 Flash 0423 is listed at $0.09 input and $0.18 output per million tokens, with a 1,048,576-token context window.

First pass over a long text file

The rate stays small even when the file is long, which is when output price starts to matter. Ask for a short output.

DeepSeek V4 Flash 0423 is listed at $0.09 input and $0.18 output per million tokens, with a 1,048,576-token context window.

The tasks are editorial illustrations, not measured benchmarks. The prices in them are read from today's feed.

Every current row

14 DeepSeek rows on the feed.
ModelKindContextInputOutputListedGuide
DeepSeek V3 Chat 163,840 $0.2574 $1.0287 26 Dec 2024 New, not yet reviewed
DeepSeek V3 0324 Chat 163,840 $0.29 $1.14 24 Mar 2025 New, not yet reviewed
DeepSeek V3.1 Chat 163,840 $0.25 $0.95 21 Aug 2025 New, not yet reviewed
R1 0528 Chat 163,840 $0.571 $2.286 28 May 2025 New, not yet reviewed
DeepSeek V3.1 Terminus Chat 163,840 $0.27 $1.00 22 Sep 2025 New, not yet reviewed
DeepSeek V3.2 Chat 163,840 $0.26 $0.38 1 Dec 2025 New, not yet reviewed
DeepSeek V3.2 Exp Chat 163,840 $0.27 $0.41 29 Sep 2025 New, not yet reviewed
DeepSeek V4 Flash 0423 Chat 1,048,576 $0.09 $0.18 24 Apr 2026 Reviewed
DeepSeek V4 Flash 0731 Chat 1,048,576 $0.0443 $0.30 31 Jul 2026 New, not yet reviewed
DeepSeek V4 Flash Vision Exp Chat 1,048,576 $0.44 $1.32 21 Aug 2026 New, not yet reviewed
DeepSeek V4 Pro 0423 Chat 1,048,576 $0.90 $1.50 24 Apr 2026 Reviewed
DeepSeek V4 Pro 0813 Chat 1,048,576 $1.32 $3.96 12 Aug 2026 New, not yet reviewed
DeepSeek V4.1 Flash Chat 1,048,576 $0.30 $1.20 10 Sep 2026 New, not yet reviewed
DeepSeek V4.1 Flash 1017 Chat 1,048,576 $0.15 $0.60 17 Oct 2026 Reviewed

Questions

Which DeepSeek model should I start with?

For most people, start with DeepSeek V4.1 Flash 1017, then drop to the next model on this page when the task is routine. Read the cheaper-model note under each example.

Are the examples benchmarks?

No. They are concrete tasks used to explain fit. The prices next to them are live figures from the feed. The judgment is a guide note.