DeepSeek V4 Pro launches with fourfold price increase as AI demand strains infrastructure capacity

DeepSeek raises some V4 prices by more than 10x as AI demand strains capacity

DeepSeek is raising its prices, to the dismay of developers, but off-peak pricing, cache discounts, and multi-model routing mean the impact is more nuanced. One of AI vendor DeepSeek's biggest selling points has been its ultra-low price point, but that party's about to end. The Chinese model provider is raising API pricing for its V4 model family by notable margins, in some cases by more than 1,100%. The increases may not be that dramatic for all, though; the company is encouraging "more flexible workload scheduling," with peak rates and half-price off-peak rates. The news was tucked into the announcement of the general availability (GA) of DeepSeek V4-Pro and upgrades to VR-Flash. The new pricing takes effect for most parts of the world on August 16.

DeepSeek's AI models are about to cost four times more - Engadget

DeepSeek's AI models are about to cost four times more Prices at off-peak hours will be half the peak-hour pricing. DeepSeek made its name offering far cheaper AI services than its pricier western competitors, but the cheap ride appears to be at an end. The company has started telling customers to prepare for significant price rises with the advent of its latest model. With the announcement of DeepSeek V4 Pro, the company is raising its API pricing fourfold. The company said it's adopting a new peak and off-peak pricing to "allocate resources more reasonably." Starting on August 16, the DeepSeek V4 Pro model will cost $3.96 for 1 million output tokens at peak hours, more than four times the current rate of $0.87. It will cost you half that amount ($1.98) during off-peak hours. Meanwhile, the DeepSeek V4 Flash model will set you back $1.32 for 1 million output tokens (from $0.28) during peak hours and $0.66 during off-peak. To be fair, the current API pricing was supposed to be just a promotion ending on May 31. DeepSeek announced that month that it was going to make the discounted prices permanent, but it ultimately decided to proceed with the price hikes. Even though DeepSeek will soon cost four times more, its V4 models are still a lot more affordable than many of its competitors. Kimi K3, a model by Chinese company Moonshot, costs $15 per 1 million output tokens. OpenAI's most advanced model, GPT-5.6 Sol, costs $30 for 1 million output tokens. However, its low cost offering, GPT-5.6 Luna, costs $1.20, less than the DeepSeek V4 Flash at peak hours.

DeepSeek officially launches V4-Pro AI model in August 2026

The Chinese AI startup released V4-Pro on its app, web, and API on Thursday, with a price increase set to follow on August 16 DeepSeek officially released its V4-Pro model on Thursday, making it available across the company's app, web interface, and API after the model had been in preview since April. The company said the general availability version, designated DeepSeek-V4-Pro-0813, focuses on agent capabilities -- tasks where AI systems use tools, execute code, and complete multi-step workflows without human intervention. Benchmark results released by DeepSeek showed the model scored 87.9 on Terminal Bench 2.1, 62.7 on DeepSWE, and 61.5 on NL2Repo, among other agent-focused tests. According to the Global Times, the model is capable of handling a context window of up to 1 million tokens and can produce outputs as long as 384,000 tokens, with the option to run in either thinking or non-thinking mode. DeepSeek also announced that the V4-Pro API has been updated to work with the OpenAI Responses API format out of the box and includes built-in support for Codex integration. Thinking effort levels for both V4-Pro and V4-Flash have been expanded to three settings -- low, high, and max -- allowing developers to match computation intensity to task complexity, the company said. A price increase for the V4 model family is set to take effect at 16:00 UTC on August 16. DeepSeek is also introducing peak and off-peak billing, with off-peak rates at half the peak-hour price, the company said. The launch comes after DeepSeek's cheaper V4-Flash model performed above expectations in independent tests conducted after its release earlier this month, in some cases outpacing the April preview of V4-Pro -- an outcome that drew attention given that Pro is positioned as the company's more capable product, according to Reuters. DeepSeek announced the price increases alongside the launch, with V4-Pro output tokens rising to $3.96 per million at peak hours from the current flat rate of $0.87 per million. Even at the new peak rates, DeepSeek's prices remain below those of some competitors -- Anthropic's Fable 5 charges $50 per million output tokens. DeepSeek had offered a 75% promotional discount on V4-Pro through May 5, making Thursday's pricing shift a notable reversal of the company's earlier strategy. DeepSeek's R1 model went viral in early 2025 and briefly gave the company a commanding position in the AI race, but competitors including Moonshot AI, Alibaba, and ByteDance have since closed the gap with their own releases. The company secured roughly $7.4 billion in its first round of outside capital in June, and Reuters reported in July that DeepSeek was in discussions for an additional round that would value it at around $74 billion.

DeepSeek V4 Pro Launches at $0.435 per Million Input Tokens

DeepSeek V4 Pro has officially launched and its arrival is already making waves across the AI industry. Priced at just $0.435 per million input tokens and $0.87 per million output tokens, it sets a new standard for affordability without sacrificing performance. According to Universe of AI, the model achieved an impressive 83.3 on the CyberGym Benchmark, narrowly surpassing Fable 5's 83.1 and outperformed it on the Automation Bench as well. These results highlight its strength in handling complex software engineering tasks while maintaining a cost-effective approach, making it a compelling option for organizations balancing technical needs with budget constraints. Explore how DeepSeek V4 Pro's pricing and performance metrics stack up against competitors and gain insight into its practical applications in fields like software development and automation. You'll also discover how its emergence, alongside models like SpaceX's Grock 4.6, is reshaping the AI landscape by prioritizing value and accessibility. Whether you're a researcher, developer, or decision-maker, this overview will help you understand the implications of these advancements and what they mean for the future of AI adoption. DeepSeek V4 Pro: Redefining Cost-Effective AI DeepSeek V4 Pro has entered the market with a clear focus on delivering exceptional performance at an accessible price point. With pricing set at $0.435 per million input tokens and $0.87 per million output tokens, it stands out as one of the most economical models available today. Despite its affordability, it delivers results that rival or even surpass premium models, making it a standout option for organizations prioritizing both performance and budget. Key performance benchmarks underscore its capabilities: * CyberGym Benchmark: Achieved a score of 83.3, narrowly surpassing the well-regarded Fable 5, which scored 83.1. * Automation Bench: Scored 31.8, outperforming Fable 5's 29.1. These results highlight its strength in deep software engineering tasks, offering a compelling solution for businesses and developers seeking a balance between cost and technical capability. For those evaluating AI models, DeepSeek V4 Pro sets a new standard for value without compromising on quality. SpaceX Grock 4.6: Innovation Meets Affordability SpaceX's Grock 4.6 is another formidable entrant in the AI market, offering competitive performance at a fraction of the cost of premium alternatives. Priced at $2 per million input tokens and $6 per million output tokens, it provides a cost-effective alternative to models like GPT 5.6 Soul, making advanced AI capabilities more accessible to a broader audience. Noteworthy features include: * Artificial Analysis Index: Matches GPT 5.6 Soul with a score of 61, demonstrating its ability to compete with top-tier models. * Advanced Output Generation: Excels in producing complex outputs, such as simulations and interactive applications, with minimal input. For developers and researchers working on interactive systems, simulations, or advanced AI-driven applications, Grock 4.6 offers a powerful yet affordable solution. Its ability to deliver high-quality results at a lower cost makes it an attractive option for projects requiring both innovation and cost efficiency. Deep dive into the latest in DeepSeek by exploring our other resources and articles. Price-to-Performance Ratio: A New Benchmark in AI The emergence of models like DeepSeek V4 Pro and Grock 4.6 underscores the growing importance of the price-to-performance ratio in the AI market. These models demonstrate that high performance no longer requires a premium price tag, creating opportunities for broader adoption and experimentation. Consider the following comparisons: * DeepSeek V4 Pro: Up to 57 times cheaper than Fable 5 while delivering nearly identical results in key benchmarks. * Grock 4.6: Offers comparable performance to GPT 5.6 Soul at a significantly reduced cost, making it a viable alternative for budget-conscious users. This shift in pricing dynamics allows organizations to achieve more with fewer resources, fostering innovation across industries. However, it also places pressure on established players to justify their higher price points, potentially driving further competition and innovation in the AI sector. Market Implications and Future Considerations The introduction of cost-effective models like DeepSeek V4 Pro and Grock 4.6 is poised to intensify competition among AI developers. Established players such as OpenAI and Anthropic may need to reevaluate their pricing strategies and innovation pipelines to maintain their market positions. For users, this increased competition translates into access to a wider range of high-performing models at reduced costs, allowing greater flexibility and experimentation. However, several questions remain regarding the long-term sustainability of these pricing strategies. Will these low-cost models maintain their performance consistency over time? Can AI labs continue to innovate while keeping prices low? These uncertainties will play a critical role in shaping the future of the AI market. Benchmark Performance and Practical Applications Both DeepSeek V4 Pro and Grock 4.6 have demonstrated strong performance across a variety of benchmarks, solidifying their positions as viable alternatives to premium models. Key results include: * DeepSeek V4 Pro: Nearly matched Fable 5 in the Terminal Bench (87.9 vs. 88) and excelled in CyberGym and Automation Bench, showcasing its versatility in software engineering tasks. * Grock 4.6: Performed competitively in GDP Val and Cursor Bench, further establishing its credibility as a cost-effective yet powerful model. These benchmarks highlight the practical applications of these models, from software development to advanced simulations. For users, staying informed about updates and real-world use cases will be essential to ensure these models continue to meet evolving project requirements. Opportunities and Challenges in the Evolving AI Landscape The release of DeepSeek V4 Pro and Grock 4.6 signals a shift in the AI industry, where cost efficiency and performance are no longer mutually exclusive. For researchers, developers and business leaders, these advancements open up new possibilities for innovation, allowing projects that were previously constrained by budget limitations. However, the industry must address challenges related to sustainability, performance consistency and market dynamics. As the AI sector continues to evolve, staying updated on new developments, benchmarks and pricing trends will be critical for making informed decisions. Whether you're exploring AI for research, development, or business applications, these models represent an exciting step forward, offering greater flexibility and potential for growth. Media Credit: Universe of AI Disclosure: Some of our articles include affiliate links. If you buy something through one of these links, Geeky Gadgets may earn an affiliate commission. Learn about our Disclosure Policy.

DeepSeek rolls out V4-Pro with improved agentic AI, DSpark decoding and Codex support

DeepSeek has launched DeepSeek-V4-Pro-0813, the official release of DeepSeek-V4-Pro that supersedes the preview version. The model is built on the DeepSeek-V4-Pro Preview architecture with a DSpark speculative decoding module and includes improvements for agentic and production workloads. DeepSeek says V4-Pro-0813 performs better than the preview version across the listed benchmarks. DeepSeek-V4-Pro: Agent and coding performance DeepSeek-V4-Pro-0813 shows higher scores than the preview model across the public and internal benchmarks provided by DeepSeek. It scores 87.9 on Terminal Bench 2.1, compared with 72.1 for V4-Pro Preview, while its scores on NL2Repo, Cybergym and DeepSWE are 61.5, 83.3 and 62.7, respectively. The model scores 74.1 on Toolathlon-Verified, 25.7 on Agents' Last Exam and 31.8 on AutomationBench (Public). On DeepSeek's internal tests, it scores 71.1 on DSBench-FullStack and 67.2 on DSBench-Hard, compared with 41.8 and 31.1 for the preview model. For the code-agent tasks in the public benchmarks, DeepSeek evaluated V4-Pro-0813 using the minimal mode of DeepSeek Harness as the agent framework, with max reasoning effort, temperature set to 1.0 and top-p set to 0.95. DeepSeek identifies DSBench-FullStack as an internal full-stack development test set and DSBench-Hard as an internal test set of difficult coding-agent problems. Flexible reasoning and API support The parameter now supports three levels: * Low: Intended for simple tasks. * High: Intended for daily agent workflows. * Max: Intended for complex tasks requiring more deliberation. The V4 lineup also adds native OpenAI Responses API support and is optimized for Codex, with a one-click setup. The API model names remain unchanged, with DeepSeek directing developers to its API documentation for setup details. The release does not include a Jinja-format chat template. Instead, DeepSeek provides a dedicated encoding folder containing Python scripts and test cases for converting OpenAI-compatible messages into input strings and parsing the model's text output. The encoding workflow also allows developers to specify the reasoning effort used for a request. DSpark speculative decoding DeepSeek-V4-Pro-0813 includes DSpark speculative decoding. Developers can use it with: * vLLM: DSpark can be enabled through the option. DeepSeek's example uses seven speculative tokens with greedy draft sampling. * SGLang: The speculative algorithm can be enabled without specifying a separate draft model path, as the target and draft weights come from the same checkpoint. * Local deployment: DeepSeek provides an inference folder with instructions for model weight conversion and interactive chat demos. Local deployment settings DeepSeek recommends the following sampling settings for local deployment: * Agentic scenarios: Temperature 1.0 and top-p 0.95. * Other scenarios: Temperature 1.0 and top-p 1.0. * High and max reasoning: Maximum output length of 384K tokens. API pricing DeepSeek is introducing separate peak and off-peak API rates with the V4 lineup. Off-peak pricing is 50% lower than peak pricing. For DeepSeek-V4-Pro, the new rates are: The new pricing takes effect at 16:00 UTC on August 16, 2026. Availability DeepSeek-V4-Pro-0813 is available on the DeepSeek app and web through Expert Mode. It is also available through the DeepSeek API, with the existing API model names retained. DeepSeek has provided API documentation for the setup process. The repository and model weights are released under the MIT License.