Current Landscape of Mental Health Conversational Agents From a Trauma-Informed Care Lens: Scoping Review

Background: Conversational agents (CAs) are increasingly used in mental health care to enhance access and engagement. However, their safe, ethical, and user-sensitive design remains a challenge. Despite growing attention to trauma-informed approaches in human-computer interaction, there is limited work on how the trauma-informed care (TIC) framework could be applied in the design of mental health CAs and no comprehensive synthesis to date. Objective: Guided by the Substance Abuse and Mental Health Services Administration’s TIC framework, this scoping review explored how TIC principles (safety; trustworthiness and transparency; collaboration and mutuality; empowerment, voice, and choice; peer support; and cultural, historical, and gender issues) are currently represented in the design and evaluation of mental health conversational agents (MHCAs) and identified gaps and opportunities to promote more trauma-informed design practices. Methods: Online databases, as well as a secondary survey of citation lists from an initial search, were used to identify English-language journal articles and conference proceedings from 2000 to 2024 that empirically evaluated an independent, web- or app-based, unassisted CA used for mental health and included concepts from TIC. Results: Our analysis included 38 publications (n=28, 73.7%, published in 2020 or later) covering 28 distinct MHCAs. Most studies used experimental methods (n=23, 60.6%) or user studies (n=11, 28.9%), with samples skewed toward female (men: mean 34.92%, SD 18.64%), young in age (mean 32.52, SD 14.6 y), and predominantly nonclinical (n=29, 76.3%). MHCAs were largely rule-based prototypes. No studies explicitly referenced the TIC framework as a guiding lens for MHCA design or evaluation. A total of 26 studies referenced terminology from TIC core principles but rarely defined them, while all 38 included language that could be linked to one or more principles. Overall, TIC-related concepts appeared most often within intervention design descriptions, qualitative assessments, or as items embedded in questionnaires evaluating broader constructs. Trustworthiness and transparency, safety, empowerment, voice and choice, and collaboration and mutuality were comparatively well addressed, while peer support and cultural, historical, and gender issues were largely absent. Design recommendations, where present, were relatively broad and emphasized secure, customizable, reliable, human-like, and context-sensitive MHCAs that offered multimodal interaction, goal setting and tracking, and transparency. Conclusions: Studies did not self-identify as using Substance Abuse and Mental Health Services Administration’s framework for TIC, making it more difficult to identify its elements. The fragmented terms, disciplines, and metrics used make it difficult to draw more systematic conclusions about the current research landscape related to TIC, but our analysis indicates TIC to be a descriptive and potentially unifying framework and provides a starting point for the explicit trauma-informed MHCA research and design.
<img src="https://jmir-production.s3.us-east-2.amazonaws.com/thumbs/c41181a042ee9ad5f9b3c8394fcddce6" />

Dissociation: Signs and Causes in Children

When people use the word dissociation, it can sound alarming. You may have seen it on social media, heard your child mention it, or noticed your child seeming “checked out” and wondered if that’s what’s happening. Dissociation can be confusing because it exists on a spectrum — from everyday experiences like daydreaming to more serious symptoms that may signal that a child is overwhelmed or struggling. The good news is that dissociation is often a temporary coping mechanism, and when it does become a problem, there are effective ways to help.

What is dissociation?

In simple terms, dissociation is a kind of mental disconnection. “When I think of dissociation, I think of there being some sort of disconnect between an individual and their sense of self, or a period of time that you later can’t recall, or feeling like you’re disconnected from your body,” says Lauren Allerhand, PsyD, a clinical psychologist at the Child Mind Institute and co-director of its DBT program.

Some kids describe dissociation as feeling spaced out, numb, or disconnected from their body or surroundings. Others say they feel like they’re watching themselves from outside their body, or that the world around them doesn’t feel real. “There’s some period of time where your normal sense of flow is disrupted,” Dr. Allerhand explains.

Is dissociation normal?

In its mildest form, dissociation is a commonplace occurrence. Kids might daydream in class, zone out during something boring like a long car trip, or feel detached when they are overwhelmed in some way. These experiences are usually not a cause for concern. “Our brains do a really good job of protecting ourselves,” Dr. Allerhand says. “Sometimes our brains develop strategies to protect us that are healthy, and other times they develop strategies that might work in short bursts but become less helpful if they happen too much.”

When dissociation happens often, or interferes with daily life, it may signal that a child is struggling with something more serious than ordinary, intermittent stress. “If it’s happening all the time, it’s less effective as a coping mechanism” because of the toll it can take when there is memory loss, confusion, and feeling disconnected to the self, she explains.

What does dissociation feel like?

Children and teens may describe dissociation differently. Some say they feel:

  • Like they’re in a dream
  • Emotionally numb
  • Detached from their body
  • Like they’re watching themselves in a movie
  • Like things around them aren’t real

“Kids might say they feel like a robot. Everything feels fake around them,” Dr. Allerhand says. “Younger children may not have the words to describe what they’re experiencing. Instead, parents might notice their child seems unusually quiet, unresponsive, or ‘not themselves.’”

Why do kids dissociate?

Dissociation is often linked to stress or overwhelming emotions — kids may dissociate when they feel unable to cope with what’s happening around them. “This could be a response to any sort of highly intense emotion or experience,” Dr. Allerhand says, such as:

  • Trauma
  • Anxiety or panic
  • Intense emotions
  • Depression
  • Major life changes
  • Overwhelming stress

“It’s another way of coping with stress or trauma,” says Tanvi Bahuguna, PsyD, a clinical psychologist at the Child Mind Institute who specializes in trauma and mood disorders. “There’s this psychological process that helps them disconnect from overwhelming pain.” Some kids dissociate during panic attacks or periods of intense anxiety. Children who have experienced significant adversity may be more likely to dissociate. These experiences can include:

  • Abuse
  • Neglect
  • Family instability (housing instability, domestic violence, addiction)
  • Loss of a family member, especially through violence or suicide

Still, experts are quick to note that dissociation doesn’t automatically mean a child has experienced trauma or has a serious disorder. “There are lots of exits on this highway before we’re at a dissociative disorder,” Dr. Allerhand says, adding that a full-blown dissociative disorder is very rare in children.

Mild vs. serious dissociation

It can be hard to recognize when a child is experiencing more serious dissociation because it doesn’t always look different from daydreaming or inattention. One key difference is distress. “Spacing out or not paying attention is not often experienced as distressing,” Dr. Allerhand says. Moderate or serious dissociation “is often somewhat distressing.” Kids who are daydreaming are still connected to themselves and their surroundings; kids who are experiencing more serious dissociation may feel cut off from their body, emotions, or reality altogether.

Using grounding techniques for dissociation

If you think your child may be dissociating, the most important thing you can do is not panic or try to get your child to “snap out of it.”  “The number one thing a parent can do is stay as calm as possible,” Dr. Bahuguna says. Speak gently, use short sentences, and reassure your child that they’re safe. Saying your child’s name and reminding them you’re there can help them reconnect.

Grounding techniques can also bring kids back into the present moment. One common method is called the 5-4-3-2-1 technique: Ask the child to name five things they can see, four things they can feel, three things they can hear, two things they can smell, one thing they can taste or imagine tasting. Other grounding strategies include:

  • Deep breathing
  • Squeezing a stress ball
  • Holding something cold
  • Gently moving the body

If you find your child often dissociates, Dr. Allerhand recommends helping them make a plan for it. During a calm moment, talk with your child about what they find helpful. “I noticed that this is happening. How can I help you when this is happening?” she suggests asking. Having a plan in advance makes it easier to respond in the moment — and in the meantime, stay nearby and make sure your child is safe until the episode passes.

When should parents seek help for a child who dissociates?

If dissociation is frequent, distressing, or associated with changes in your child’s functioning, seeking professional support is appropriate. “If something dissociative happens, and there’s a really big change in your child’s functioning, then I would be concerned,” Dr. Allerhand says.

Signs it may be time to reach out include:

  • Memory gaps after the episode
  • Noticeable personality changes
  • Difficulty at school
  • Withdrawal from friends or activities
  • Significant distress or confusion

A good place to start would be talking to your pediatrician, who may refer you to a mental health professional. “If your child is displaying behaviors that seem out of the ordinary, you should trust your instincts,” Dr. Allerhand says.

How to identify dissociation

To determine whether a child is dissociating, a mental health professional gathers information from multiple sources, including parents, the child, and sometimes teachers, asking about the child’s behaviors, history, and any recent stressors or changes in behavior.

“The first thing would be a structured diagnostic interview with a qualified clinician,” Dr. Allerhand explains. “Parents bring the history and describe the behavior, and then the clinician meets with the child.” Clinicians also consider whether dissociation might be a symptom of another condition, such as post-traumatic stress disorder, borderline personality disorder, anxiety (especially panic disorder), and depression.

“It’s really gathering history, meeting the child, observing the child, and figuring out what this cluster of behaviors leads to,” she says. It’s more frequent to find that dissociation is a result of another disorder than an actual dissociative disorder.

How is dissociation treated?

Treatment depends on what’s driving the dissociation. If trauma is involved, therapy may focus on helping the child process difficult experiences and build coping skills. Evidence-based approaches include trauma-focused cognitive behavioral therapy (TF-CBT) and eye movement desensitization and reprocessing (EMDR).

If anxiety or emotional overwhelm is the primary cause, treatment may focus on emotion regulation, grounding techniques, and identifying triggers and early warning signs. Therapy, such as dialectical behavior therapy (DBT), typically involves both children and parents, helping families recognize patterns and respond in supportive ways.

For more severe or persistent dissociation, treatment may happen in phases — beginning with safety and stabilization, then skill-building, and eventually, when appropriate, processing difficult experiences. “The goal is helping the child learn to cope with their experience and stay in their body,” Dr. Allerhand says.

What are dissociative disorders?

In children and teens, dissociation is usually a symptom of another condition. But in cases of very serious early trauma, abuse, or neglect, it can progress into a full-blown disorder. There are a number of dissociative disorders, including:

  • Dissociative identity disorder (what was once called multiple personality disorder) involves two or more distinct personality states and gaps in memory and is typically linked to significant early trauma. Parents who search online may find alarming information, but Dr. Allerhand says this condition is very rare in kids.
  • Dissociative amnesia involves gaps in memory that can’t be explained by ordinary forgetfulness — such as not remembering important personal information or periods of time — and is often associated with stressful or traumatic experiences.
  • Depersonalization/derealization disorder involves feeling detached from oneself, as though watching yourself from outside your body, or feeling that the world around you isn’t real.

These disorders sometimes attract media attention, but they are extremely rare in children. What’s important for parents to know is that if you see dissociative behavior in a child, it’s most likely a normal coping mechanism for a child experiencing some stress or intense emotion. If it persists, is causing distress, or is interfering with a child’s life, it’s time to consult a pediatrician or mental health professional. Identifying what might be causing the behavior is the first step to getting appropriate treatment.

Frequently Asked Questions

What is dissociation?

Dissociation is a mental disconnection from your thoughts, feelings, body, or surroundings. Kids may feel spaced out, numb, or like they’re watching themselves from the outside, as if the world doesn’t feel real.

What are common symptoms of dissociation?

Common signs include feeling detached from the body, emotionally numb, or like you’re in a dream. Some kids seem unusually quiet or “not themselves,” while others have trouble recalling what happened during that time.

What causes dissociation?

Dissociation is often a response to stress, anxiety, or overwhelming emotions. It can also be linked to trauma, major life changes (such as the sudden loss of a family member), or intense feelings the child doesn’t yet know how to manage.

How can you stop dissociating?

Grounding techniques can help bring you back to the present moment, like naming what you see, hear, and feel, or focusing on breathing. Having a plan for what you will do the next time can make it easier to manage when it happens.

The post Dissociation: Signs and Causes in Children appeared first on Child Mind Institute.

Real-world effectiveness of medication-assisted treatment and psychotherapy for opioid use disorder: a national multi–health care organization analysis

BackgroundHarm reduction strategies for opioid use disorder (OUD) emphasize pragmatic, evidence-based approaches that reduce overdose risk, relapse, and other adverse outcomes without requiring abstinence. Medication for opioid use disorder (MOUD) and structured psychotherapy represent core harm-reduction modalities, yet their real-world comparative effectiveness, alone and in combination, remains underexplored at scale.MethodsA retrospective cohort study was conducted using the TriNetX Research Network, comprising de-identified electronic health records from 112 U.S. health systems. 18,047 adults aged 18–45 were identified with a diagnosis of opioid dependence (ICD-10 F11.20) between 2016 and 2025. Subjects were assigned to eight mutually exclusive treatment cohorts: no treatment (Cohort 1); buprenorphine alone (Cohort 2); methadone alone (Cohort 3); psychotherapy alone (30 minutes (Cohort 4), 45 minutes (Cohort 5), or 60 minutes (Cohort 6)); buprenorphine + psychotherapy (Cohort 7); and methadone + psychotherapy (Cohort 8), with combination treatments defined within a ±30-day window. Cox proportional hazards models estimated adjusted hazard ratios (aHRs) for remission (F11.21, F11.11) within 12 months.ResultsBuprenorphine (aHR = 2.33; 95% CI: 1.85–2.94), methadone (aHR = 2.50; 95% CI: 2.05–3.04), and psychotherapy (30 min: aHR = 2.18; 45 min: aHR = 2.38) were each independently associated with significantly higher remission compared to no treatment. The combination of buprenorphine + psychotherapy yielded the strongest effect (aHR = 5.26; 95% CI: 2.68–10.32). Anxiety diagnoses and gabapentinoid prescriptions were positively associated with remission; benzodiazepine co-prescription was negatively associated.ConclusionsIn this first national-scale, multi–health-care-organization analysis, both pharmacologic and psychosocial harm-reduction interventions were independently associated with improved OUD remission, with additive benefit when integrated. These findings underscore the value of embedding comprehensive, multimodal harm-reduction services within routine care and support policies promoting equitable access to both MOUD and behavioral health supports across diverse health systems.

Scoping review of therapeutic approaches among individuals with secondary exercise addiction

Secondary exercise addiction shows high comorbidity with eating and body image disorders. Despite its substantial impact on physical and mental health and daily functioning, evidence on effective therapeutic interventions remains limited. The aim of this scoping review was to identify and describe therapeutic interventions applied to adult individuals with secondary exercise addiction. This review followed the PRISMA Sc-R guidelines and covered the years 2002–2024. Ultimately, five studies were included (four randomized controlled trials and one quasi-experimental study). Three studies applied psychotherapeutic interventions based on cognitive-behavioral models (Cognitive Behavioral Therapy, Lifestyle, Exercise, Attitudes, and Relationships Program, Physical Exercise and Dietary Therapy), while two integrated physical or nutritional components. A secondary analysis published in 2024 based on the LEAP trial dataset was identified but not treated as an independent study to avoid duplication. EBSCOhost, Web of Science, PubMed, and Google Scholar were searched from January to May 2025 using terms related to exercise addiction, exercise abuse, psychotherapy, intervention, and treatment. English-language studies were eligible if they described an intervention with at least one treated group with pre- and post-test measures; the participants of the study were adult patients suffering from eating disorders and exercise addiction (the therapy programs involved one inpatient and four outpatient treatments) and therapeutic intervention was carried out with outcomes based on exercise addiction level data. Four out of five included studies reported improvements in variables related to compulsivity, although these did not always imply a reduction in the amount of exercise, indicating that qualitative changes may be more relevant. Longer interventions showed more consistent effects, but even brief treatments generated positive changes in non-clinical populations. The examination of the research revealed a gap in studies addressing interventions for those with secondary exercise addiction, especially highlighting the need for randomized controlled trials (RCTs) with proper randomization methods.

Trump administration warns against using federal dollars on fentanyl test strips

The Trump administration is doubling down on its opposition to harm reduction services for people who use illicit drugs. 

In an open letter on April 24, the federal agency overseeing addiction and mental health policy warned its grantees against using federal funds to buy harm reduction supplies including sterile syringes and pipes, or to distribute test strips for common drug supply adulterants like fentanyl, xylazine, and medetomidine. 

Read the rest…

How adolescent cannabis use reshapes the developing brain — a systematic review

Background and hypothesisCannabis use initiation during adolescence has increased globally, raising concerns about neurodevelopmental consequences during this critical period when the brain undergoes extensive remodeling in cannabinoid receptor-rich regions.Study designThis systematic review examines neurodevelopmental consequences of adolescent cannabis use, focusing on structural brain changes, cognitive impacts, addiction vulnerability, and long-term outcomes. We searched PubMed, EMBASE, PsycINFO, and Web of Science (2000-2025) for studies examining cannabis effects in adolescent populations. Following PRISMA guidelines, two reviewers screened 3,421 records and assessed 156 full-text articles, including studies with neuroimaging, cognitive assessments, or longitudinal follow-up.Study resultsThirty-six studies involving 8,432 participants met criteria: 23 longitudinal cohorts (62.2%), 8 cross-sectional (22.2%), 4 RCTs (11.1%), and 1 case-control study (2.8%). Neuroimaging revealed dose-dependent alterations including reduced prefrontal cortical and hippocampal/amygdala volumes, accelerated cortical thinning in longitudinal studies, and impaired white matter connectivity correlating with initiation age. Cognitive findings were mixed — some showed persistent deficits after prolonged abstinence in adolescent-onset users, others found no effects after controlling for confounders. Epidemiological studies consistently showed elevated addiction risk (ORs 3.9–7.2) in adolescents versus adults. Long-term associations included educational difficulties, mental health problems, and functional impairment, though causal relationships remained unclear.ConclusionsAdolescent cannabis use associates with structural brain changes, elevated addiction risk, and variable cognitive effects, suggesting greater vulnerability versus adult-onset use. However, methodological limitations including confounders, heterogeneous definitions, and observational designs limit causal inference. Findings support age-specific prevention and specialized interventions while highlighting needs for rigorous longitudinal research establishing causality.Systematic review registrationhttps://www.crd.york.ac.uk/prospero/, identifierCRD420251165329.

The Download: DeepSeek’s latest AI breakthrough, and the race to build world models

This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology.

Three reasons why DeepSeek’s new model matters

On Friday, Chinese AI firm DeepSeek released a preview of V4, its long-awaited new flagship model. Notably, the model can process much longer prompts than its last generation, thanks to a new design that handles large amounts of text more efficiently.

While the model remains open source, its performance matches leading closed-source rivals from Anthropic, OpenAI, and Google. It is also DeepSeek’s first release optimized Huawei’s Ascend chips—a key test of China’s dependence on Nvidia.

Here are three ways V4 could shake up AI.

—Caiwei Chen

The rise of world models

AI systems have already gained impressive mastery over the digital world, but the physical world remains humanity’s domain. As it turns out, building an AI that composes novels or code apps is far easier than developing one to fold laundry or navigate city streets. To bridge this gap, many researchers believe you need something called a world model.

Proponents like Stanford professor Fei-Fei Li and AMI Labs founder Yann LeCun argue these models can overcome the well-known limitations of LLMs—and realize AI’s promise for robotics. Find out why they’ve brought world models to the forefront of the field.

—Grace Huckins

World models are on our list of the 10 Things That Matter in AI Right Now, our essential guide to what’s really worth your attention in the field.

Subscribers can watch an exclusive roundtable unveiling the technologies and trends on the list, with analysis from MIT Technology Review’s AI reporter Grace Huckins and executive editors Amy Nordrum and Niall Firth.

The must-reads

I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology.

1 China has blocked Meta’s $2 billion acquisition of AI startup Manus
Regulators cited national security grounds. (WSJ $)
+ Beijing called the deal a “conspiratorial” attempt to hollow out its tech base. (FT $)
+ The country is tightening its grip on AI firms that try to leave. (TechCrunch)
+ The decision escalates China’s AI rivalry with the US. (Bloomberg $)
+ But there will be no winners in their competition. (MIT Technology Review)

2 Google is investing up to $40 billion in Anthropic
In a deal valuing the AI firm at $350 billion. (CNBC)
+ The funding will support the firm’s growing computing needs. (TechCrunch)
+ Anthropic and OpenAI are fighting for compute capacity. (Axios)

3 President Trump just fired the entire National Science Board
The NSF has played a crucial role in developing technology. (The Verge)
+ The move heightens fears over political interference in US science. (Nature)

4 Conspiracy theories about the Washington shooting are proliferating online
Over 300,000 posts appeared on X using the keyword “staged.” (NYT $)
+ The theories are also swirling on Bluesky and Instagram. (Wired)

5 The AI compute crunch is starting to hit the broader economy.
It’s affecting jobs, gadgets, and electricity prices. (404 Media)
+ The AI compute explosion is the tech story of our time. (MIT Technology Review)

6 Elon Musk says a new banking tool brings X close to a “super app”
He’s pledged to launch the tool this month. (Bloomberg)

7 AI optimism is surging across Asia while US sentiment cools
The divide could shape where adoption happens fastest. (Rest of World)

8 Apple is tying its new CEO’s ascent to its first foldable iPhone
It wants to build the buzz around John Ternus. (Gizmodo

9 Twelve firms are developing the Golden Dome’s space-based interceptors
They’ve won contracts worth up to $3.2 billion. (Ars Technica)

10 NASA has shared promising results from Artemis II
The spacecraft and rocket fared well. (Engadget)

Quote of the day

“Getting out the truth and establishing facts and reliable information takes time. But our audiences really don’t have that kind of patience.”

—Amanda Crawford, associate professor at the University of Connecticut, tells the NYT why conspiracy theories are gaining traction online.

One More Thing

MIRIAM MARTINCIC


Welcome to Kenya’s Great Carbon Valley: a bold new gamble to fight climate change

Kenya’s Great Rift Valley is home to five geothermal power stations, which harness clouds of steam to generate about a quarter of the country’s electricity. But some of the energy escapes into the atmosphere, while even more remains underground for lack of demand. That’s what brought Octavia Carbon here.

Last year, the startup began harnessing some of that excess energy to remove CO2 from the air. The company says the method is efficient, affordable, and—crucially—scalable. But the project also faces fierce opposition. 

Read the full story on the future of Kenya’s “Great Carbon Valley.”


—Diana Kruzman

We can still have nice things

A place for comfort, fun and distraction to brighten up your day. (Got any ideas? Drop me a line.)

+ Fred Again’s Tiny Desk Concert is a masterclass in intimate performance.
+ Here’s a delightful look at how we’re all linked through geography and shared heritage.
+ Take a short, peaceful break to watch Tokyo’s cherry blossoms from a bird’s eye view.
+ There’s something oddly satisfying about watching an industrial shredder turn everyday items into confetti.

Three reasons why DeepSeek’s new model matters

On Friday, Chinese AI firm DeepSeek released a preview of V4, its long-awaited new flagship model. Notably, the model can process much longer prompts than its last generation, thanks to a new design that helps it handle large amounts of text more efficiently. Like DeepSeek’s previous models, V4 is open source, meaning it is available for anyone to download, use, and modify.

V4 marks DeepSeek’s most significant release since R1, the reasoning model it launched in January 2025. R1, which was trained on limited computing resources, stunned the global AI industry with its strong performance and efficiency, turning DeepSeek from a little-known research team into China’s best-known AI company almost overnight. It also helped set off a wave of open-weight model releases from other Chinese AI firms. 

DeepSeek has kept a relatively low profile since then—but earlier this month, it effectively teased V4’s release when it added “expert” and “flash” modes to the online version of its model, prompting speculation that the updates were tied to a bigger upcoming release.

While the company has become a powerful symbol of China’s AI ambitions, its big return to cutting-edge frontier models comes after months of scrutiny—including major personnel departures, delays to previous model launches, and growing scrutiny from both the US and Chinese governments. 

So, will V4 shake the AI field the way R1 did? Almost certainly not, but here are three big reasons why this release matters.

1. It breaks new ground for an open-source model.

As with R1 before it, DeepSeek claims that V4’s performance rivals the best models available at a fraction of the price. This is great news for developers and for companies using the tech, because it means they can access frontier AI capabilities on their own terms, and without worrying about skyrocketing costs.

The new model comes in two versions, both of which are available on DeepSeek’s website and in its app, with API access also open to developers. V4-Pro is a larger model built for coding and complex agent tasks, and V4-Flash is a smaller version designed to be faster and cheaper to run. Both versions offer reasoning modes, in which the model can carefully parse a user’s prompt and show each step as it works through the problem.

For V4-Pro, DeepSeek charges $1.74 per million input tokens and $3.48 per million output tokens, a fraction of the cost of comparable models from OpenAI and Anthropic. V4-Flash is even cheaper, at about $0.14 per million input tokens and about $0.28 per million output tokens, making it one of the cheapest top-tier models available. This would make it a very appealing model to build applications on.

In terms of performance, V4 is, perhaps unsurprisingly, a huge jump from R1—and it seems to be a strong alternative to just about all the latest big AI models. On the major benchmarks, according to results shared by the company, DeepSeek V4-Pro competes with leading closed-source models, matching the performance of Anthropic’s Claude-Opus-4.6, OpenAI’s GPT-5.4, and Google’s Gemini-3.1. And compared to other open-source models, such as Alibaba’s Qwen-3.5 or Z.ai’s GLM-5.1, DeepSeek V4 exceeds them all on coding, math, and STEM problems, making it one of the strongest open-source models ever released. 

DeepSeek also says that V4-Pro now ranks among the strongest open-source models on benchmarks for agentic coding tasks and performs well on other tests that measure ability to carry out multistep problems. Its writing ability and world knowledge also leads the field, according to benchmarking results shared by the company. 

In a technical report released alongside the model, DeepSeek shared results from an internal survey of 85 experienced developers: More than 90% included V4-Pro among their top model choices for coding tasks.

DeepSeek says it has specifically optimized V4 for popular agent frameworks such as Claude Code, OpenClaw, and CodeBuddy.

2. It delivers on a new approach to memory efficiency.

One of the key innovations of V4 is its long context window—the amount of text the model can process at once. Both versions can handle 1 million tokens, which is large enough to fit all three volumes of The Lord of the Rings and The Hobbit combined. The company says this context window size is now the default across all DeepSeek services and it matches what is offered by cutting-edge versions of models like Gemini and Claude. 

But it’s important to know not just that DeepSeek has made this leap, but how it did so. V4 makes significant architectural changes to the company’s former models—especially in the attention mechanism, which is the feature of AI models that helps them understand each part of a prompt in relation to the rest. As the prompt text gets longer, these comparisons become much more costly, making attention one of the main bottlenecks for long-context models.

DeepSeek’s innovation was to make the model more selective about what it pays attention to. Instead of treating all earlier text as equally important, V4 compresses older information and focuses on the parts most likely to matter in the present moment, while still keeping nearby text in full so it does not miss important details. 

DeepSeek says this sharply reduces the cost of using long context. In a 1-million-token context, V4-Pro uses only 27% of the computing power required by its previous model, V3.2, while cutting memory use to 10%. The reduction in V4-Flash is even larger, using just 10% of the computing power and 7% of the memory. In practice, this could make it cheaper to build tools that need to work across huge amounts of material, such as an AI coding assistant that can read an entire codebase or a research agent that can analyze a long archive of documents without constantly forgetting what came before.

DeepSeek’s interest in long context windows didn’t start with V4. Over the past year and a half, the company has quietly published a series of papers on how AI models “remember” information, experimenting with compression and mathematical techniques to extend what AI models could realistically handle.

3. It marks the first steps on the hard road away from Nvidia.

V4 is DeepSeek’s first model optimized for domestic Chinese chips, such as Huawei’s Ascend—a move that has turned the launch into something of a test of whether China’s homegrown AI industry can begin to loosen its dependence on US chip giant Nvidia. 

This was largely expected, since The Information reported earlier this month that DeepSeek did not give American chipmakers like Nvidia and AMD early access to V4, though prerelease access is common to allow chipmakers to optimize support of the new model ahead of a launch. Instead, the company reportedly gave early access only to Chinese chipmakers. 

On Friday, Huawei said its Ascend supernode products, based on the Ascend 950 series, would support DeepSeek V4. This means that companies and individuals who want to run their own modified version of Deepseek V4 will be able to use Huawei chips easily.

Reuters previously reported that Chinese government officials recommended that DeepSeek integrate Huawei chips in its training process. And this pressure fits a broader pattern in China’s industrial policy: Strategic sectors are often pushed, and sometimes effectively required, to align with national self-reliance goals. But there’s a particular urgency when it comes to AI. Since 2022, US export controls have cut Chinese firms off from Nvidia’s most powerful chips, and they later also restricted access to downgraded China-market versions. Beijing’s response has been to accelerate the push for a domestic AI stack, from chips to software frameworks to data centers.

Chinese authorities have reportedly been pushing data centers and public computing projects to use more domestic chips, including through reported bans on foreign-made chips, sourcing quotas, and requirements to pair Nvidia chips with Chinese alternatives from companies such as Huawei and Cambricon. 

Still, replacing Nvidia is not as simple as swapping one chip for another. Nvidia’s advantage lies not only in its chips, but in the software ecosystem developers have spent years building around them. Moving to Huawei’s Ascend chips means adapting model code, rebuilding tools, and proving that systems built around those chips are stable enough for serious use.

To be clear, DeepSeek does not appear to have fully moved beyond Nvidia. The company’s technical report reveals that it is using Chinese chips to run the model for inference, or when someone asks the model to complete a task. But Liu Zhiyuan, a computer science professor at Tsinghua University, told MIT Technology Review that DeepSeek appears to have adapted only part of V4’s training process for Chinese chips. The report does not say whether some key long-context features were adapted to domestic chips, so Liu says V4 may still have been trained mainly on Nvidia chips. Multiple sources who spoke on the condition of anonymity, due to political sensitivity around these issues, told MIT Technology Review that Chinese chips still don’t perform as well as Nvidia chips but are better suited for inference than training.

DeepSeek is also tying the future costs of V4 to this hardware shift. The company says V4-Pro prices could fall significantly after Huawei’s Ascend 950 supernodes begin shipping at scale in the second half of this year. 

If that works, V4 could be an early sign that China is successfully building a parallel AI infrastructure.