Doha [Qatar], April 26: DeepSeek's long-awaited flagship V4 model has fallen short of its domestic and US rivals, according to a new analysis, as the Chinese artificial intelligence firm struggles to replicate the market-shaking success of its earlier R1 release.
The company's most advanced system, V4 Pro, ranked second among the world's leading open-source models, behind Beijing-based Moonshot AI's Kimi K2.6, benchmark firm Artificial Analysis said in a report on Friday.
While V4 Pro marked a clear improvement on its predecessor, V3.2, it still lagged top competitors.
V4 Pro scored 52 on the Artificial Analysis Intelligence Index, compared with 54 for Kimi K2.6, released earlier this week. By contrast, leading closed-source models from the US - OpenAI's GPT-5.5, Anthropic's Claude Opus and Google's Gemini 3.1 Pro - scored 60, 57 and 57, respectively.
The results highlight the challenges facing DeepSeek as China races to narrow the AI gap with the US, amid intensifying competition at home and abroad as well as ongoing constraints on computing power.
Still, analysts noted that V4 delivered meaningful technical progress.
DeepSeek's V4 was "impressive" for coming close to state-of-the-art performance, featuring an efficient one-million-token context window and the ability to run on Huawei Technologies' Ascend 950PR AI chips, said Kyle Chan, a research fellow at the Brookings Institution, in a post on X on Friday.
A context window refers to the amount of information an AI model can process in a single pass. DeepSeek's previous flagship model had a context window of 128,000 tokens.
In a report on Saturday, research firm SemiAnalysis said DeepSeek's 90 per cent reduction in KV cache in a one-million-token context setting was "far more impactful than Google's TurboQuant paper last month".
The model is also notable for its compatibility with domestic hardware. Shortly after V4's release on Friday, Huawei Technologies said its Ascend chip range and supernode systems would provide "full support" for running the model in inference.
But questions remain over how the model was trained. Brookings Institution fellow Kyle Chan said it was "equally notable" that DeepSeek made no mention of using Chinese chips during training, even as the model continued to trail US frontier systems.
That view was echoed by Council on Foreign Relations senior fellow Chris McGuire who said the release did little to shift the broader picture of US leadership in AI. He estimated the US remained about seven months ahead, adding that the absence of detail on training costs or hardware suggested V4 may have relied on restricted Nvidia Blackwell chips.
DeepSeek did not immediately respond to a request for comment.
Market reaction was also more muted than during the debut of its earlier R1 reasoning model. While shares of Chinese chipmakers rallied on Friday following news of V4 and its integration with Huawei hardware, the release failed to trigger the kind of global shock seen last year.
When R1 launched, it wiped hundreds of billions of dollars from US equity markets, with Nvidia shares plunging 17 per cent in a single day. On Friday, however, Nvidia stock rose 4.32 per cent.
Artificial Analysis also flagged potential drawbacks in the new model. Despite gains in knowledge benchmarks, V4 Pro and its lighter V4 Flash variant recorded hallucination rates of 94 per cent and 96 per cent, respectively.
The firm further noted that V4 Pro is now more expensive than rival open-source models, including Kimi K2.6 and Zhipu AI's GLM-5.1, as well as DeepSeek's own V3.2. Even so, it remained significantly cheaper than leading closed-source systems, according to the benchmark firm.
SemiAnalysis called DeepSeek's V4 "an exceptional engineering release" that was "just behind" the frontier. While its capabilities were not at the leading edge, it said the model could serve as a low-cost alternative to US closed-source systems.
Source: Qatar Tribune
