DeepSeek V4 Pro 0813 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 Pro 0813

DeepSeek V4 Pro 0813 is documented as a packaged model artifact on NVIDIA's NIM for Large Language Models and Vision Language Models platform, with a dedicated "Get Started" page maintained across NIM versions 2.0.10 through 3.0.0, indicating active platform-side support for the identifier over an extended release window. Independent third-party coverage of the same model identifier exists in the form of a MindStudio blog post titled "DeepSeek-V4-Pro-0813: What's New and How It Stacks Up," published shortly after the model's release.

No creator-side technical report, model card, or release note from DeepSeek is present in the supplied evidence, so architecture lineage, parameter count, training data composition, benchmark results, and qualitative reasoning behavior cannot be stated from the available sources. The model's practical fit — its suitability for long-context reasoning, tool-driven workflows, or structured-output generation — is likewise unsupported by the excerpts on hand and is therefore left undescribed rather than speculated.

Meliousdeepseek-v4-pro-0813deepseek-thinking

Open provider page Provider docs

Quick Info

Provider: Melious
Model key: deepseek-v4-pro-0813
Release date: Aug 12, 2026
Last updated: Aug 22, 2026
Input modalities
Output modalities
Capabilities

Cost

Input token cost: $1.1592
Output token cost: $3.4776

Limits

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

Transparent token rates

Compare DeepSeek V4 Pro 0813 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 Pro 0813

Melious

CoverageRelease Notes

DeepSeek moved its flagship V4-Pro out of preview on 13 August 2026 with the DeepSeek-V4-Pro-0813 build, keeping the existing one-million-token context window and parameter count while upgrading behavior rather than architecture. Callers were switched to the new checkpoint without code changes, and the model name remai Operationally, both DeepSeek-V4-Pro-0813 and DeepSeek-V4-Flash-0731 gain a selectable reasoning effort with low, high, and max settings, replacing the previous single fixed thinking budget, and DeepSeek recommends "high" for everyday agent work. On DeepSeek's own evaluation harness in minimal mode, V4-Pro-0813 scores 8

Melious

CoverageBenchmark

The AI Release Tracker page documents DeepSeek-V4-Pro-0813 as released on Thursday, 13 August 2026, 13 days after DeepSeek-V4-Flash-0731, and lists the benchmark suite the model is evaluated against, including BullshitBench v2, SWE-Bench Verified, Terminal-Bench 2.1, Toolathlon-Verified, BrowseComp, and CyberGym. The e API pricing is quoted per million tokens and verified against api-docs.deepseek.com on 18 August 2026. Off-peak rates are $0.66 input, $0.022 cached input, and $1.98 output; peak rates (01:00–04:00 and 06:00–10:00 UTC) are $1.32 input, $0.044 cached input, and $3.96 output. The page also surfaces the model's one-millio