DeepSeek V4.1 Flash
DeepSeek V4.1 Flash is a reasoning model from DeepSeek released on Sep 10, 2026. It reads text and images and writes text, with a 1M token context window and up to 393K tokens per reply.
Show moreShow less
It costs $0.15 per million input tokens and $0.60 per million output tokens. Cached input bills at $0.003, 98% less than fresh input.
It supports tool calling, structured JSON output and prompt caching, and its weights are open, so you can host it yourself.
56 providers serve it, including Deep Infra, Baseten and Fireworks AI. The lowest input price is $0.01 per million tokens at CoralBricks.
On the Artificial Analysis Intelligence Index it scores 39.5, which places it at number 17 among the 43 models I track with independent scores.
- Modalities
- In / out price
- $0.15 / $0.60 per 1M
- Context
- 1M
- Max output
- 393K
- Released
- Sep 10, 2026
- Knowledge cutoff
- May 2025
- Tool calling
- Structured output
- Reasoning
- Prompt caching
- Open weights
- Free to use
DeepSeek V4.1 Flash pricing
USD, as listed on Oct 15, 2026. Most bills are input heavy, so the cached input rate matters as much as the headline price.
| Input | $0.15 / 1M tokens | |
|---|---|---|
| Output | $0.60 / 1M tokens | |
| Cached input (read) | $0.003 / 1M tokens | 98% off input |
DeepSeek V4.1 Flash providers
56 providers serve it, including Deep Infra, Baseten and Fireworks AI. The lowest input price is $0.01 per million tokens at CoralBricks. Each host sets its own price and limits, so the cheapest one is not always the one with the longest context.
| Provider | Input /M | Output /M | Cache read /M | Context | Max output |
|---|---|---|---|---|---|
| DeepSeek First party | $0.15 | $0.60 | $0.003 | 1M | 393K |
| Deep Infra | $0.20 | $0.60 | $0.006 | 1.05M | 384K |
| Baseten | $0.30 | $1.20 | $0.03 | 1.05M | 33K |
| Fireworks AI | $0.30 | $1.20 | $0.006 | 1M | 384K |
| Hugging Face | $0.30 | $1.20 | – | 1.05M | 384K |
| Nebius Token Factory | $0.30 | $1.20 | $0.30 | 1.05M | 1.05M |
| Novita AI | $0.30 | $1.20 | $0.006 | 1.05M | 393K |
| Together AI | $0.30 | $1.20 | $0.006 | 1.05M | 384K |
| Vercel AI Gateway | $0.30 | $1.20 | $0.007 | 1.05M | 33K |
| Azure | n/a | n/a | – | 1M | 128K |
| Nvidia | n/a | n/a | – | 1M | 384K |
| CoralBricks | $0.01 | $1.20 | $0 | 1.05M | 131K |
Showing 12 of 56 providers: the lab first, then the major clouds, then the cheapest resellers.
What DeepSeek V4.1 Flash costs per month
Pick a workload close to yours or type your own volumes.
Independent benchmarks
Measured by third parties, not reported by DeepSeek. Each links to its source and shows the date I checked it.
A composite of 10 independent evaluations (agents 30%, coding 20%, scientific reasoning 20%, general 30%), run by Artificial Analysis on each model, version 4.3.2.
Artificial Analysis, checked Oct 10, 2026
Human preference: people compare two anonymous answers and pick the better one. Ratings from the Text Arena overall board, style control on.
Arena (formerly LMArena), checked Oct 10, 2026
Abstract reasoning on visual puzzles the model has not seen, semi private set. Hard to memorise, so it tracks reasoning more than knowledge.
ARC Prize, checked Oct 10, 2026
Median output tokens per second measured by Artificial Analysis on the provider API, for the same model setting as its index score.
Artificial Analysis, checked Oct 10, 2026
Benchmarks DeepSeek reports (12)
- gpqa-diamond (GPQA Diamond (Pass@1), Max reasoning effort)90.9%
- hle-no-tools (HLE (Pass@1), Max reasoning effort, full set)36.8%
- hle-with-tools (HLE w/ tools (Pass@1), Max reasoning effort)63.9%
- terminal-bench-2.1 (Terminal-Bench 2.1 (Pass@1), DeepSeek Harness Minimal mode, 1M context, Max reasoning effort)90.6%
- terminal-bench-4.0 (Terminal-Bench 4.0 (Pass@1), DeepSeek Harness Minimal mode, Max reasoning effort)31.2%
- deepswe-v1.1 (DeepSWE v1.1 (Resolved), mini-SWE harness, Max reasoning effort)74.2%
- nl2repo-bench (NL2Repo-Bench (Score), Max reasoning effort)64
- codeforces (Codeforces (Rating), Max reasoning effort)3471
- matharena-apex (MathArena Apex (Pass@1), Max reasoning effort)65.6%
- automationbench (AutomationBench (Pass@1), Max reasoning effort)54.8%
- cybergym (CyberGym (Pass@1), Max reasoning effort)88.1%
- mmmu-pro (MMMU-Pro (EM), 4-shot, BASE model (DeepSeek-V4.1-Flash-Base))56.5%
Labs pick their own settings and harnesses, so compare these only within one lab.
Supported request parameters
What you can set when you call DeepSeek V4.1 Flash through an OpenAI compatible API.
- Tools
tools - Structured outputs
structured_outputs - Reasoning
reasoning - Temperature
temperature
Alternatives to DeepSeek V4.1 Flash
Models that take the same inputs and call tools, closest in price, cheaper ones first.
- Index 11.3
Reads text and images. Tool calling, structured output, reasoning. Served by 17 providers.
by mistralaiMar 16, 2026262K context$0.15/M input tokens$0.60/M output tokens17 providers
- Index 41.8
Reads text, images, files and video. Tool calling, structured output, reasoning. Served by 65 providers.
by z-aiAug 21, 20261M context$0.15/M input tokens$0.50/M output tokens65 providers
- Index 39.8
Reads text, images and video. Tool calling, structured output, reasoning. Served by 23 providers.
by qwenAug 26, 20261M context$0.15/M input tokens$0.47/M output tokens23 providers
- Index 14.7
Reads text and images. Tool calling, structured output, reasoning. Served by 42 providers.
by googleApr 2, 2026262K context$0.14/M input tokens$0.40/M output tokens42 providers
- Index 43.4
Reads text, images and files. Tool calling, structured output, reasoning. Served by 26 providers.
by anthropicOct 7, 20261M context$0.10/M input tokens$0.50/M output tokens26 providers
- Index 38.1
Reads text, images and files. Tool calling, structured output, reasoning. Served by 27 providers.
by openaiSep 22, 20261.05M context$0.10/M input tokens$0.50/M output tokens27 providers
More from DeepSeek
- 1.05M context
Reads text and images. Tool calling, reasoning.
by deepseekOct 13, 2026$0.15/M input tokens$0.60/M output tokens
- 1.05M context
Reads text and images. Tool calling, structured output, reasoning. Served by 14 providers.
by deepseekAug 21, 2026$0.44/M input tokens$1.32/M output tokens14 providers
- 1M context
Reads text. Tool calling, structured output, reasoning.
by deepseekAug 2, 2026$0.05/M input tokens$0.16/M output tokens
- 1.05M context
Reads text. Tool calling, structured output, reasoning. Served by 48 providers.
by deepseekJul 31, 2026$0.14/M input tokens$0.28/M output tokens48 providers
- 1.05M context
Reads text. Tool calling, structured output, reasoning.
by deepseekMay 3, 2026$0.48/M input tokens$1.98/M output tokens
- 1.05M context
Reads text. Tool calling, structured output, reasoning. Served by 46 providers.
by deepseekApr 24, 2026$0.14/M input tokens$0.28/M output tokens46 providers
DeepSeek V4.1 Flash questions
How much does the DeepSeek V4.1 Flash API cost?
$0.15 per million input tokens and $0.60 per million output tokens, with cached input at $0.003. The estimator on this page turns that into a monthly bill for your own workload.
What is the context window of DeepSeek V4.1 Flash?
1,000,000 tokens of input, and it can write up to 393,216 tokens in one reply. As a rough guide, a million tokens is around 750,000 English words.
Which providers serve DeepSeek V4.1 Flash?
56 providers serve it, including Deep Infra, Baseten and Fireworks AI. The lowest input price is $0.01 per million tokens at CoralBricks. Prices differ by host, so the Providers table on this page lists each one with its input and output price.
Does DeepSeek V4.1 Flash support tool calling?
Yes. It accepts tool definitions and can return JSON that matches a schema, which is what an agent needs to call your APIs.
Is DeepSeek V4.1 Flash open source?
Its weights are published, so you can run it on your own hardware or a cloud of your choice. Read the license before commercial use.
When was DeepSeek V4.1 Flash released?
Sep 10, 2026, with training data up to May 2025.
How good is DeepSeek V4.1 Flash compared with other models?
It scores 39.5 on the Artificial Analysis Intelligence Index and 1475 on the Arena text leaderboard. Benchmarks are a shortlist, not a verdict: test the top two or three candidates on your own tasks before you commit.
Catalog data from models.dev (MIT licensed), refreshed daily, last on Oct 15, 2026. Independent scores from my hand checked rankings, sources linked above. The summary is written from these facts.
Building on DeepSeek V4.1 Flash?
I test your real tasks on it and the two closest alternatives, then ship the agent on whichever wins on cost and accuracy.
Book Meeting