DeepSeek V4.1 Flash 1017: API and Pricing

DeepSeek V4.1 Flash 1017 is the October 17 update to DeepSeek V4.1 Flash. Morph lists it as morph-dsv41flash-1017 with the same per-token prices as morph-dsv41flash. The two names share a serving pool and have separate billing records. Check the live model catalog for readiness before sending production traffic.

What changed in DeepSeek V4.1 Flash 1017?

DeepSeek reports improvements in reasoning, coding, agent tasks, and visual benchmarks, alongside shorter outputs and task execution times. Its model card reports the following results; these are publisher measurements, not independent Morph evaluations.

Source: DeepSeek's 1017 model card. DeepSeek used Claude Code for Terminal-Bench 4.0 and Mini-SWE-V2 for DeepSWE v1.1. Harness differences matter when comparing these scores with other reports.

DeepSeek V4.1 Flash 1017 API pricing

Morph charges $0.12 per million uncached input tokens, $0.008 per million cached input tokens, and $0.8 per million output tokens. Both public IDs use these list rates. Customer-specific discounts are tracked separately for each ID.

For example, a request with 100,000 uncached input tokens and 10,000 output tokens costs $0.020 at list rates. If all those input tokens are cache hits, it costs $0.0088. Actual cost follows measured token usage. See the full Morph pricing table.

Model IDs and availability

Use morph-dsv41flash-1017 to identify requests under the new billing entry. Existing integrations can keep morph-dsv41flash. The old name remains supported, but it is not a guarantee of the older checkpoint: both names route to the same underlying deployment.

The catalog identifies the new weights as deepseek-ai/DeepSeek-V4.1-Flash-1017. During rollout, its is_ready field stays false until serving activation is complete. Use the entry's current readiness, pricing, and context fields when configuring an integration.

Call the model with an OpenAI-compatible client

After the new entry is ready, set the base URL to https://api.morphllm.com/v1 and the model to morph-dsv41flash-1017. Keep your API key in an environment variable. The example below uses the OpenAI Python client.

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["MORPH_API_KEY"],
    base_url="https://api.morphllm.com/v1",
)

response = client.chat.completions.create(
    model="morph-dsv41flash-1017",
    messages=[{
        "role": "user",
        "content": "Review this function for bugs and suggest a fix: ...",
    }],
)
print(response.choices[0].message.content)

Create an API key or read the model API documentation for integration details. For direct OpenRouter use, its model name is deepseek/deepseek-v4.1-flash-1017; provider pricing and availability can differ from Morph's direct API.

How to evaluate the update

Compare saved baseline results with a fresh run of the same tasks after activation. Calling both Morph IDs at the same time compares billing names, not two independent checkpoints. Record task completion, total billed tokens, elapsed time, and tool failures with the same prompts and harness. A higher benchmark score alone does not establish lower cost for your workload.

For the earlier model family and its API history, read the DeepSeek V4 guide. For available alternatives, context windows, and current prices, see all Morph models.