DeepSeek V4 Flash 0731 by Melious | AI model information

$10 off the fastest DeepSeek V4.1 Flash, Kimi K3 and GLM 5.3$10 off the fastest DeepSeek V4.1 Flash, Kimi K3 and GLM 5.3

Model details

DeepSeek V4 Flash 0731

DeepSeek V4 Flash 0731 is positioned as an efficiency-oriented variant in the V4 lineup, built on a 284B-parameter CSA plus HCA backbone where only 13B parameters activate per token. This sparse design lets the model deliver throughput suitable for agentic and long-context pipelines without the compute footprint of its larger V4 siblings. The July 31, 2026 checkpoint represents a re-post-training pass that emphasizes agent reliability rather than altering the underlying architecture, with the DSpark speculative-decoding module attached to accelerate generation in production deployments.

According to the community release write-up, this build outperforms the V4-Pro preview across the agentic benchmarks DeepSeek publishes, making it a strong default for developers building agents, evaluation harnesses, or long-context workflows on top of the V4 stack. MIT-licensed open weights shipped alongside the checkpoint, lowering the barrier for self-hosting and fine-tuning. Native Responses API and Codex support further smooth integration for tool-using applications, while the combination of agent-tuned post-training and the speculative decoder positions the model as a practical, cost-conscious choice for production agent systems.

Meliousdeepseek-v4-flash-0731deepseek-flash

Open provider page Provider docs

Quick Info

Provider: Melious
Model key: deepseek-v4-flash-0731
Release date: Jul 31, 2026
Last updated: Jul 31, 2026
Knowledge cutoff: 2025-05
Input modalities
Output modalities
Capabilities

Cost

Input token cost: $0.11592
Output token cost: $0.2898

Limits

Output tokens: 384,000 tokens
Context window: 1,000,000 tokens

Transparent token rates

Compare DeepSeek V4 Flash 0731 pricing

Rates are shown per one million tokens. Combined means one million input plus one million output tokens.

Browse this family

Latest news about DeepSeek V4 Flash 0731

Melious

Coverage

A community thread on the NVIDIA DGX Spark / GB10 forum, titled "Deepseek-v4-Flash 0731 GGUF (NEW model)" and dated July 31, 2026, treats DeepSeek-V4-Flash-0731 as a newly released open-weight model from DeepSeek. According to specs quoted from Unsloth's page within the post, the 0731 build is an update to the Flash va Because the page is a user-to-user DGX Spark discussion rather than a primary DeepSeek announcement, the strongest contributions are the explicit naming of DeepSeek-V4-Flash-0731 and the restatement of its parameter counts, context length, and intended use cases. Readers looking for a release announcement should still

Melious

Coverage

DeepSeek's official API changelog (api-docs.deepseek.com) is the only first-party source in the candidate set and contains a September 10, 2026 entry titled "DeepSeek-V4.1-Flash Release." Alongside the V4.1-Flash announcement, the changelog explicitly states that "the previous-generation models V4 Flash and V4 Flash Vi The relevance to the subject is narrow but high-impact: the changelog does not provide release notes or benchmark numbers specific to the July 31, 2026 build, but it is the authoritative source for the deprecation status of the deepseek-v4-flash model name as of September 10, 2026. For developers and users targeting De