DeepSeek V4.1 Flash 1017 API: now on FastMetal

DeepSeek V4.1 Flash 1017 (deepseek-v4.1-flash-1017) is now available on FastMetal, through the API and in chat. It is the October 17, 2026 snapshot of DeepSeek V4.1 Flash. The undated deepseek-v4.1-flash stays in the catalog at the same price, so you can run both side by side before switching.

What changed from V4.1 Flash

DeepSeek's model card calls 1017 a minor update with gains in reasoning, coding, agentic tasks and visual benchmarks. A few rows from the card:

DeepSeek also says the snapshot finishes tasks faster and writes shorter outputs. We did not see that on our host: sent the same 14 prompts, the two snapshots produced about the same number of output tokens, and 1017 usually took longer to answer. Neither reasons by default unless you ask it to.

Capabilities

• Image and PDF input. We confirmed both with real files before launch: an image described correctly, and a one-page PDF read back word for word.
• Tool calling and structured output. A forced tool call and a strict json_schema response both worked on real requests.
• 1M-token context. Served at 1,048,576 tokens.
• Open weights. Released under the MIT license on Hugging Face. We serve it from a host that does not train on prompts.

Pricing

$0.211 per 1M input tokens and $0.633 per 1M output tokens, the same as V4.1 Flash. Cached input is billed at $0.00633 per 1M. Billing is prepaid, so you can never spend more than you have loaded. Current rates are on the model page and pricing.

Calling it

FastMetal's API is OpenAI-compatible. Set the model to deepseek-v4.1-flash-1017:

curl https://api.fastmetal.ai/v1/chat/completions \
  -H "Authorization: Bearer $FASTMETAL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v4.1-flash-1017",
    "messages": [
      {"role": "user", "content": "Find the bug in this Python function and show a fixed version."}
    ]
  }'

Already on deepseek-v4.1-flash? Changing the model name is the whole migration. To try it without code, pick DeepSeek V4.1 Flash 1017 in chat.

Compare real answers first

The model page shows DeepSeek V4.1 Flash 1017's actual answers to our 43 shared prompts: summarization, bug fixing, translation and more. Put them next to DeepSeek V4.1 Flash to decide whether to switch.

Browse the full model catalog.