Short answer
DeepSeek V4.1 Flash 1017 is the Flash model update DeepSeek released on October 17, 2026. DeepSeek says it scores higher, finishes tasks faster and writes shorter outputs than the previous version. The model name deepseek-flash in DeepSeek's API now points to it. Official pricing depends on the time of day: off-peak it is $0.15 per million input tokens and $0.60 per million output tokens, and peak hours cost twice as much. On TeamoRouter it is available as deepseek-flash-1017 at about $0.113 input / $0.451 output off-peak at the time of writing, roughly 75% of the official price.
What changed
From DeepSeek's release note:
- Better scores, faster, shorter outputs. DeepSeek calls it an improvement in both performance and cost.
- Tuned for agent tools. DeepSeek names DeepSeek Harness, Tencent CodeBuddy and WorkBuddy.
- Whole projects from scratch. Reinforcement learning with self-feedback improved end-to-end project work; the demos include 3D modeling and a browser racing game.
- Open weights on Hugging Face and ModelScope.
Specs
| DeepSeek V4.1 Flash 1017 | |
|---|---|
| DeepSeek model name | deepseek-flash |
| TeamoRouter model ID | deepseek-flash-1017 (pinned to this version), deepseek-flash (follows DeepSeek's latest Flash) |
| Released | October 17, 2026 |
| Context window | 1M tokens |
| Max output | 384K tokens |
| Thinking mode | Thinking and non-thinking; thinking is the default |
| Tool calls, JSON output | Supported |
The old names are retired. According to DeepSeek, the models behind deepseek-v4-flash and deepseek-v4-flash-vision-exp have been retired; requests to those names in DeepSeek's API are now served by V4.1 Flash 1017 and billed at the Flash price. For new code, use deepseek-flash-1017.
Pricing and peak hours
DeepSeek's price per 1 million tokens:
| Off-peak | Peak | |
|---|---|---|
| Input (cache miss) | $0.15 | $0.30 |
| Input (cache hit) | $0.003 | $0.006 |
| Output | $0.60 | $1.20 |
Peak hours are 01:00–04:00 and 06:00–10:00 UTC, Monday to Friday, except Chinese public holidays. Everything else, including weekends, is off-peak.
TeamoRouter uses the same peak and off-peak hours. deepseek-flash-1017 off-peak at the time of writing:
| DeepSeek off-peak | TeamoRouter off-peak | |
|---|---|---|
| Input | $0.15 | ≈ $0.113 |
| Cache hit | $0.003 | ≈ $0.0023 |
| Output | $0.60 | ≈ $0.451 |
Both double at peak. TeamoRouter prices move with upstream; the current numbers are on the pricing page. If a batch job can wait, running it off-peak halves the bill.
Use it in Claude Code
V4.1 Flash works over the Anthropic protocol, so Claude Code needs three environment variables. Get the key in the TeamoRouter dashboard:
export ANTHROPIC_BASE_URL=https://api.teamorouter.com
export ANTHROPIC_AUTH_TOKEN="your TeamoRouter API key"
export ANTHROPIC_MODEL=deepseek-flash-1017
For a hard step, switch to Claude in the same session with /model claude-opus-5-5; the key stays the same. Full setup: Claude Code guide.
Call it with the OpenAI SDK
from openai import OpenAI
client = OpenAI(base_url="https://api.teamorouter.com/v1", api_key="YOUR_KEY")
resp = client.chat.completions.create(
model="deepseek-flash-1017",
messages=[{"role": "user", "content": "Write a Python function that checks whether a number is prime"}],
)
print(resp.choices[0].message.content)
Full reference: API integration.
Paying for usage
No subscription: you top up a balance from $5 and pay only for the tokens you use, by card or with USDT, USDC or ETH from any exchange app (how to pay with crypto). The same key also works for Claude, GPT and Gemini.
FAQ
Are V4.1 Flash and V4.1 Flash 1017 the same model? 1017 is the October 17 update of V4.1 Flash. DeepSeek's deepseek-flash points to it now; on TeamoRouter, use deepseek-flash-1017 to stay on this exact version.
I still call deepseek-v4-flash. Should I change it? Yes. DeepSeek has retired the model behind that name; use deepseek-flash-1017.
When is DeepSeek V5 coming? As of October 22, 2026, DeepSeek has not announced a release date, and there is no V5 entry in its changelog.
How do I avoid peak prices? Peak hours only fall on weekday working hours in China. Evenings, weekends and Chinese public holidays are billed at the off-peak rate.