DeepSeek-V4-Pro-0813 review, pricing and limits

On verified benchmarks, DeepSeek-V4-Pro-0813 scores 80.6% on SWE-bench Verified, 90.1% on GPQA Diamond, 87.5% on MMLU-Pro, and 93.5% pass@1 on LiveCodeBench. That SWE-bench score matches Gemini-3.1-Pro's agentic coding result, per Unite.AI's August 2026 coverage. DeepSeek's own change log for the GA build reports 42.7 on Humanity's Last Exam without tools and 87.9 on Terminal-Bench 2.1. Artificial Analysis, reading its v5.0.1 Intelligence Index on November 9, 2026, calls the model below average among comparable open-weight models and measures output at 68 tokens per second, with a long wait for the first answer token because the model reasons before it replies.

The model ships a 1,000,000-token context window with a 384,000-token maximum output, and the two share the same budget: a 700,000-token prompt leaves under 300,000 tokens for generation. Long context is handled by a hybrid attention design across a 61-layer stack, alternating Compressed Sparse Attention with Heavily Compressed Attention, which cuts single-token inference compute to 27% of DeepSeek-V3.2's cost at maximum context length and reduces KV cache size to roughly 10%.

DeepSeek-V4-Pro-0813 is text-only: it accepts and produces text, with no native image, audio, or video input, despite pre-launch reporting that expected multimodal training to ship. It supports tool calls and structured JSON output, and the 0813 build added native OpenAI Responses API compatibility aimed specifically at Codex-style agent integrations. Callers choose between non-think, think-high, and think-max reasoning-effort tiers per request.

DeepSeek's pricing page, read on November 9, 2026, charges double during weekday peak hours (01:00 to 04:00 and 06:00 to 10:00 UTC) compared with every other hour, including weekends and Chinese public holidays. Today's rates sit well above the $0.435 and $0.87 launch rate carried over from the April preview, so budgets built on older coverage run low. The cost FAQ below has the full breakdown.

Access runs through DeepSeek's own OpenAI-compatible API, capped at 500 concurrent connections per account, plus Microsoft Azure AI Foundry and third-party hosts including Fireworks AI, Together AI, DeepInfra, SiliconFlow, Nebius, Novita, and Baseten, plus the OpenRouter aggregator. AWS Bedrock and Google Vertex AI listings were called "coming soon" at the April launch but were not confirmed live as of this listing. Self-hosting the open weights (FP4+FP8 mixed and FP8 mixed Safetensors formats) requires a multi-node GPU cluster even at reduced quantization; exact VRAM figures are detailed in the limitations below.

Post-training combines supervised fine-tuning with GRPO reinforcement learning and on-policy distillation across the three reasoning tiers, per DeepSeek's Hugging Face model card. Independent red-teaming from FAR.AI found the safeguards brittle: a public roleplay jailbreak template first circulated for an earlier DeepSeek model raised the StrongREJECT jailbreak success rate from 0.6% to 77.8% on V4-Pro with no modification, and agentic-misuse template attacks cut refusal below 1%. Refusal is also framing-dependent: binary yes/no prompts triggered an 88.6% refusal rate, but the same requests phrased as free-form generation saw refusal collapse to about 1%, with harmful persuasion attempted on 85-95% of the non-controversial harmful topics FAR.AI tested.

DeepSeek-V4-Pro-0813 suits cost-sensitive teams running agentic coding pipelines or long-document synthesis who want frontier-adjacent benchmark scores without frontier US-lab pricing, and infra teams that specifically want to self-host open weights under a permissive MIT license. Skip it for consumer-facing deployment without an added moderation layer, given the jailbreak findings above, and for workloads needing native vision or audio input, where GPT-5.4 or Gemini-3.1-Pro are the safer default.

DeepSeek pretrained the model on more than 32 trillion tokens, over double V3's 14.8 trillion, with deliberate emphasis on long documents and agentic execution traces, but has not published an official training-data cutoff date for V4. Its core infrastructure runs out of Hangzhou, China, a different data-residency profile than US- or EU-hosted alternatives; compliance certification details are covered in the FAQ below.

The 0813 build closed a preview window that began April 24, 2026 with the joint V4-Pro and V4-Flash launch. V4-Flash graduated to general availability first, on July 31, 2026, with V4-Pro following two weeks later. DeepSeek discontinued the legacy deepseek-chat and deepseek-reasoner API model names on July 24, 2026, forcing callers onto the versioned deepseek-v4-pro endpoint.

The lineup has moved on since the GA build. DeepSeek replaced V4-Flash with DeepSeek V4.1 Flash, updated to DeepSeek V4.1 Flash 1017 on October 17, 2026. That sibling is the cheaper tier, with native vision and a 2,500-request concurrency limit against this model's 500. DeepSeek said on September 10, 2026 that it would keep serving V4 Pro past September 14, so this build remains on sale, and the earlier DeepSeek V4 Pro preview record covers the April launch.