What changed
DeepSeek published V4.1 Flash 1017 on October 17, 2026. The model card calls it a minor update to V4.1 Flash with improvements in reasoning, coding, agent tasks and visual benchmarks, and says it finishes tasks faster with shorter outputs.
In DeepSeek's own table, the 1017 snapshot scores 77.0 on DeepSWE v1.1 (up from 74.2), 42.3 on Terminal-Bench 4.0 (up from 31.2) and 60.5 on Automation-Bench (up from 54.8). DeepSeek ran Terminal-Bench with the Claude Code harness and DeepSWE with a Mini-SWE-V2 harness.
Who can use it and when
The model is live on DeepSeek's website, app and API. In the API it is served under the model name deepseek-flash, with a 1M-token context, up to 384K output tokens, thinking and non-thinking modes, tool calls, JSON output and vision. Requests to the legacy names deepseek-v4-flash and deepseek-v4-flash-vision-exp are now also served by the 1017 model and billed at the Flash price.
The weights are published on Hugging Face and ModelScope under the MIT license. The release has no Jinja chat template; DeepSeek provides a Python reference encoder and its deepseek-recipe libraries for the prompt format.
Why it matters
Because the API name did not change, anyone using deepseek-flash switched to the new snapshot automatically. That can improve results, but it can also change behaviour in a working pipeline, so rerun your own evaluation set and compare output length and tool calls with earlier logs.
The model does not lead every test in DeepSeek's table: Claude Opus 5 scores higher on Terminal-Bench 4.0, ProgramBench, NL2Repo-Bench and HLE. The argument is price: on DeepSeek's DeepSWE chart, the 1017 snapshot reaches 77.0% at an average cost per task of about $0.20.
Access and limitations
DeepSeek's API pricing for deepseek-flash per million tokens is $0.30 input, $0.006 cached input and $1.20 output at peak times, and half that off-peak ($0.15, $0.003 and $0.60). Peak hours are 01:00 to 04:00 and 06:00 to 10:00 UTC on weekdays, excluding Chinese public holidays. Other providers, such as OpenRouter, set their own prices for the snapshot.
Server-side search and fetch calls from DeepSeek Harness, Claude Code and Codex are free for now; DeepSeek says billing at $0.75 per 1,000 requests starts on October 31, 2026 at 16:00 UTC. The benchmark results are self-reported, and AppFor.ai has not tested the model.
Go to the source
Sources & further reading
Based on the official documentation below. Features and access can change; check the provider before making a decision.
Was this page useful?
Help us make our coverage more useful. This is feedback on the page, not a rating of the tool.
Go a little further