--- title: "Tencent Hy4 Preview Review: 1M Context & Cost" canonical: https://wavect.io/blog/tencent-hy4-preview-coding-agent-review/ language: en description: "Tencent Hy4 preview review: verify its 1M context, coding benchmarks, API price, WorkBuddy test and self-hosting needs before a team pilot." image: "https://wavect.io/img/blog/headers/header_tencent-hy4-preview-coding-agent-review.png" --- [**Back**](/blog/overview/) [![Kevin Riedl](/img/team/kevin.webp)](/team/kevin-riedl/) [Kevin Riedl](/team/kevin-riedl/) https://linkedin.com/in/wsdt 10 min read · 1 Sep 2026 Last reviewed September 1, 2026 [**Next**](/blog/pagelm-self-hosted-ai-study-platform/) # Tencent Hy4 Preview Review: Is the 1M-Context Coding Model Worth a Pilot? TL;DR Tencent Hy4 preview is a credible pilot candidate for long-running coding agents, not an automatic replacement for a proven production model. Tencent reports 770B backbone parameters, 49B active per token, a 1M-token context window and an Apache 2.0 release. Its internal 203-task evaluation narrowly beat GLM-5.3 and Kimi K3, while public benchmark records show strong coding results, but those figures do not prove codebase retention or reliability on your stack. Hosted access is the practical starting point because official self-hosting guidance targets 8 B300 or 16 B200 GPUs. Run representative repository tasks, measure accepted changes rather than raw tokens, verify tool use and security, and treat the preview's overthinking and over-verification as measurable cost risks. **Yes, Tencent Hy4 preview is worth a controlled coding-agent pilot if your work is limited by long context, multi-step tool use, or repeated repository retrieval.** It is not yet a safe default for every production workflow. The model is enormous, self-hosting is a datacenter project, and Tencent explicitly labels this an early version that can reason and verify longer than necessary. Tencent released Hy4 preview on 28 August 2026 with 770 billion backbone parameters, 49 billion active parameters per token, and a context window above one million tokens. The [official Tencent announcement](https://www.tencent.com/tencent-releases-and-open-sources-tencent-hy4-preview/) also lists WorkBuddy, CodeBuddy, Tencent Cloud TokenHub, and OpenRouter as access routes. Launch access in WorkBuddy and CodeBuddy was announced as free for two weeks. This review was checked on **1 September 2026**. It owns one narrow search intent: whether a software team should pilot Hy4 preview for long-horizon coding work. For a market-wide model shortlist, use our separate [open-weight LLM comparison](/blog/open-weight-llm-comparison-2026/). For the architecture choice between retrieval, tuning, and large context, use the [RAG vs fine-tuning vs long-context guide](/blog/rag-vs-finetune-vs-longcontext-2026/). ## Hy4 preview facts in 60 seconds | Question | Verified answer | Buyer implication | | --- | --- | --- | | What is it? | A 770B-total, 49B-active Mixture-of-Experts language model | Large capacity with sparse activation, not a 49B checkpoint | | How much context? | 1,048,576 tokens in the serving recipe | Enough capacity for very large inputs, but capacity is not proof of recall | | Is it open? | Weights and code under Apache 2.0 | Commercial use and self-hosting are possible, subject to a normal legal review | | Can it call tools? | Yes, official vLLM and SGLang recipes include reasoning and tool parsers | Agent integration is supported, but your harness still owns permissions and recovery | | What does the official API cost? | $0.834 input, $2.501 output, and $0.042 cache hit per 1M tokens | Long prompts can be affordable when cache reuse is real | | Can a team self-host it? | Yes, with official recipes targeting 8 B300 or 16 B200 GPUs | Hosted evaluation should come before infrastructure procurement | | Main known limitation? | Overthinking and over-verification on complex work | Measure time, output volume, and cost per accepted change | ## What is Tencent Hy4 preview? **Hy4 preview is Tencent Hy Team's flagship sparse Mixture-of-Experts model for coding, office work, analysis, games, and scientific tasks.** The [official Hugging Face model card](https://huggingface.co/tencent/Hy4-preview) documents 78 backbone layers. The first uses a dense feed-forward network; the other 77 contain 256 routed experts and one shared expert. Each token activates eight routed experts plus the shared expert. The 770B figure covers the backbone. A native multi-token-prediction layer adds another 10B total parameters and 0.7B active parameters for speculative decoding, which explains why some model listings show 780B. The architecture combines Gated DeepSeek Sparse Attention, cross-layer IndexCache reuse, and identity Hyper-Connections. The practical goal is to make long inputs and sustained reasoning more efficient without activating the full model for every token. The weights, an FP8 variant, fine-tuning code, deployment guidance, and license are available in the [official Tencent Hy4 preview repository](https://github.com/Tencent-Hunyuan/Hy4-preview). That is meaningful openness. It does not make a 770B model a laptop download. ## What did our WorkBuddy coding test show? We asked Hy4 preview inside [WorkBuddy](https://www.workbuddy.ai/) to build a fully interactive AI model benchmark dashboard. The useful result was not that it produced HTML. It also wrote the core logic, built headless harnesses, verified JavaScript against a stub DOM, and checked filters plus cost-simulation edge cases before delivery. **That is promising evidence of workflow completion, not an independent benchmark.** One successful dashboard does not establish a general success rate, security level, or economic advantage. It does reveal a good pilot hypothesis: Hy4 may be valuable when an agent must retain requirements, implementation state, test evidence, and tool output across a longer run. ## How strong are the Hy4 preview benchmarks? Tencent reports that 163 internal experts blindly rated outputs on 203 engineering tasks. Hy4 averaged 2.99 out of 4, narrowly above GLM-5.3 at 2.92 and Kimi K3 at 2.94. Against GLM-5.3, Tencent reports 46.8% wins, 12.8% ties, and 40.4% losses. Against Kimi K3, the result was 51.2% wins, 7.9% ties, and 40.9% losses. | Signal | Published result | What it supports | What it does not prove | | --- | --- | --- | --- | | Tencent expert evaluation | 2.99/4 across 203 engineering tasks | Competitive performance on Tencent's real-work task set | Independent superiority or your repository success rate | | SWE-bench Pro record | 65.7 | Strong issue-resolution capability under that harness | Production maintainability, latency, or safety | | DeepSWE record | 64.3 | Useful coding-agent potential | Reliable million-token codebase understanding | | SWE-bench Multilingual record | 82.9 | Promising multilingual code work | Equal performance across your languages and frameworks | Use the results to justify an evaluation, not a migration. Vendor and public benchmark runs differ in harnesses, tools, context, compute, retry policy, and scoring. Hy4 is also a preview, so providers can change serving behavior while the model remains under active development. ## Does a 1M-token window mean the model remembers a whole codebase? **No. A one-million-token context window is an input capacity, not a memory or accuracy guarantee.** It can reduce destructive chunking and let the model inspect broader relationships in one run. It does not guarantee that a detail near the middle will influence the correct edit, that generated code will stay consistent, or that every token is worth paying to send. The [LongCodeBench research paper](https://arxiv.org/abs/2505.07897) found that long-context coding performance can fall sharply as input grows, even when a model technically supports the window. Hy4 was released after that study and was not evaluated in it. The study supports the testing method, not a claim about Hy4's score. For production work, provide a repository map, task boundaries, changed-file hints, acceptance tests, and concise prior evidence. Our [coding-agent context guide](/blog/ai-coding-agents-context-not-intelligence/) explains why structured context usually beats an undifferentiated repository dump. ## Hosted API or self-hosted Hy4? **Most teams should start with WorkBuddy, CodeBuddy, TokenHub, OpenRouter, or another reviewed hosted route.** Tencent's published API price is $0.834 per million input tokens, $2.501 per million output tokens, and $0.042 per million cache-hit tokens. Confirm the live rate, region, retention, throughput, context limit, and tool support with the route you actually buy. The [official vLLM recipe](https://recipes.vllm.ai/tencent/Hy4-preview) lists the full deployment at 16 NVIDIA B200 or 8 B300 GPUs. Official vLLM and SGLang examples expose an OpenAI-compatible endpoint, use the FP8 model, and enable dedicated Hy4 reasoning and tool-call parsers. This is viable infrastructure for a team that already operates distributed inference. It is not a sensible first experiment for a normal product team. | Route | Choose it when | Main risk | Decision gate | | --- | --- | --- | --- | | WorkBuddy or CodeBuddy | You want the fastest end-to-end product trial | Less control over harness details and telemetry | Can it complete your real workflow with inspectable evidence? | | Hosted API | You need integration, measured usage, and tool calls | Provider-specific limits, data terms, and serving variance | Does cost per accepted task beat the current route? | | Self-hosted FP8 | You have high stable volume, data-control needs, and distributed-GPU operations | Capital, utilization, networking, upgrades, and on-call ownership | Does measured TCO beat a contracted API at required reliability? | Use our [local model versus API break-even model](/blog/local-models-vs-apis-break-even-eu-2026/) before turning open weights into a hardware purchase. The right denominator is accepted tasks, not theoretical tokens. ## A two-week Hy4 coding-agent pilot 1. **Freeze 20 to 30 tasks.** Include bug fixes, repository navigation, feature work, test repair, long documents, and tool failures from real codebases. 2. **Keep the harness comparable.** Use the same repository commits, instructions, tools, time limits, and acceptance tests for Hy4 and the current model. 3. **Test context in bands.** Run compact context, curated large context, and a near-window stress case. Do not assume more input is better. 4. **Capture the complete cost.** Record uncached input, cache hits, output, retries, elapsed time, and active human review minutes. 5. **Exercise tools and recovery.** Inject failed commands, malformed output, a stale file, and an unavailable dependency. Verify that the agent stops or recovers predictably. 6. **Review security boundaries.** Classify source code and data, restrict credentials and network access, redact logs, and require approval for production or destructive actions. 7. **Decide from accepted outcomes.** Expand only if Hy4 improves first-pass acceptance, cost per accepted change, review time, or a strategic data-control requirement. ## What should a commercial evaluation measure? - **First-pass acceptance rate:** tasks accepted without a second implementation attempt. - **Cost per accepted change:** model cost, retries, infrastructure, and review divided by accepted changes. - **Long-context retrieval accuracy:** evidence found and used correctly at several input positions. - **Tool completion rate:** valid calls, correct arguments, error recovery, and clean stop behavior. - **Verification quality:** relevant tests run, failures interpreted correctly, and unsupported claims avoided. - **P50 and P95 lead time:** the preview's overthinking may hide inside tail latency. - **Security exceptions:** denied actions, secret exposure, unapproved network access, and manual overrides. If one model cannot win every lane, keep the routing decision separate from this review. Our [multi-model coding-agent buying guide](/blog/multi-model-ai-coding-agent-stack-2026/) shows how to assign planning, implementation, and independent verification without hardcoding one vendor into the workflow. ## Recommendation **Pilot Hy4 preview when long-running engineering work is a measured bottleneck and you can compare it against a real baseline.** Use hosted access first. Keep repository context structured, instrument cache behavior, and make tests rather than the model's self-confidence the acceptance gate. Wait if your tasks fit comfortably inside a smaller context, your codebase lacks reliable tests, or nobody owns provider risk and agent permissions. Self-host only when a validated workload, data-control requirement, and stable utilization justify 8 to 16 datacenter GPUs plus the people required to operate them. ## Primary sources and freshness boundary Architecture, access routes, pricing, the 203-task internal evaluation, and the two-week launch offer come from Tencent's 28 August announcement. Detailed specifications, public evaluation records, limitations, and Apache 2.0 licensing come from the Tencent model card and repository. Hardware guidance comes from the vLLM recipe. LongCodeBench provides a general method for questioning million-token coding claims; it did not test Hy4. Wavect did not reproduce Tencent's benchmark suite. Product access, provider prices, privacy terms, and serving limits can change after the 1 September 2026 review date. ## Frequently Asked Questions About Tencent Hy4 Preview ### What is Tencent Hy4 preview? Hy4 preview is a Tencent Hy Team Mixture-of-Experts language model aimed at coding, agent workflows, office work, analysis, games, and science. Its backbone has 770 billion total parameters, with 49 billion activated per token, plus a separate 10-billion-parameter MTP layer for speculative decoding. ### Does Hy4 preview really support one million tokens? Yes. The model card lists a 1M context length and the vLLM recipe specifies 1,048,576 tokens. That is capacity, not proof that the model recalls and uses every code detail accurately at maximum length. Test retrieval and task success across several context bands. ### Is Hy4 preview open source? Tencent publishes the model weights, FP8 weights, code, fine-tuning guidance, and deployment recipes under Apache 2.0. Teams should still review the license, third-party components, model behavior, and their own regulatory duties before production use. ### How much does the Hy4 preview API cost? Tencent's 28 August announcement lists 0.834 US dollars per million input tokens, 2.501 dollars per million output tokens, and 0.042 dollars per million cache-hit tokens. Confirm the live price and conditions with the provider route you plan to use. ### Can Hy4 preview run on a laptop or one GPU? Not the official full deployment. The vLLM recipe lists 16 B200 or 8 B300 GPUs for the FP8 model. Community quantizations may reduce memory needs, but they require separate quality, performance, and license verification. ### Is Hy4 preview better than GLM-5.3 or Kimi K3? Tencent's blind internal evaluation put Hy4 slightly ahead on its 203 engineering tasks. The margins were small and the evaluation was vendor-run. Choose from a controlled test on your repositories, tools, languages, limits, and acceptance criteria. ### What is the biggest Hy4 preview risk for coding agents? The largest practical risks are treating context capacity as reliable memory, paying for unnecessary reasoning and verification, trusting tool calls without external controls, and choosing datacenter-scale self-hosting before a hosted pilot proves business value. ## Final thoughts Hy4 preview deserves attention because Tencent combined an unusually large open model, a million-token context window, agent-oriented tool support, and aggressive hosted pricing. The WorkBuddy dashboard test suggests that this stack can carry implementation and verification through a meaningful full-stack task. The buying decision still belongs to your evidence. Test real repositories, compare accepted changes, price the entire workflow, and keep tool permissions plus external tests in control. If Hy4 wins there, scale the route. If it only wins the specification table, keep watching the preview mature. ## You may also like.. [**Local Models vs APIs: The EU Break-Even Guide** Calculate when utilization, hardware, energy, operations, and data control justify self-hosted inference.](/blog/local-models-vs-apis-break-even-eu-2026/) [**Multi-Model AI Coding Agent Stack** Route planning, implementation, and verification by measured capability instead of model hype.](/blog/multi-model-ai-coding-agent-stack-2026/) Models and infrastructure ## Continue through this cluster Model selection, inference economics, local deployment, compression and serving architecture. [Start with the cornerstone**Self-Hosting LLMs in the EU: When Open Weights Actually Pay Off**](/blog/self-hosting-llms-eu-cost/) - [PageLM Review: Self-Hosting and Commercial Use](/blog/pagelm-self-hosted-ai-study-platform/) - [M6 Mac mini vs M5 Mac Studio for Local AI: Buyer's Guide](/blog/mac-mini-m6-vs-mac-studio-m5-local-ai/) - [FreeToken AI Review: Run Frontier MoE Models on Consumer GPUs?](/blog/freetoken-ai-inference-engine-review/) - [Darkbloom AI Review: Private Inference on Idle Macs](/blog/darkbloom-ai-private-inference-mac/) - [Ox Alpha Free AI Model: Setup, Privacy and Buyer Guide](/blog/ox-alpha-free-ai-model-guide-2026/) [**Back**](/blog/overview/) [![Kevin Riedl](/img/team/kevin.webp)](/team/kevin-riedl/) [Kevin Riedl](/team/kevin-riedl/) https://linkedin.com/in/wsdt 10 min read · 1 Sep 2026 Last reviewed September 1, 2026 [**Next**](/blog/pagelm-self-hosted-ai-study-platform/) ## Structured Data ```json { "@context": "https://schema.org", "@graph": [ { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ], "employee": [ { "@id": "https://wavect.io/team/kevin-riedl/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Kevin Riedl", "url": "https://wavect.io/team/kevin-riedl/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } }, { "@id": "https://wavect.io/team/christof-jori/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Christof Jori", "url": "https://wavect.io/team/christof-jori/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } } ], "founder": [ { "@id": "https://wavect.io/team/kevin-riedl/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Kevin Riedl", "url": "https://wavect.io/team/kevin-riedl/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } }, { "@id": "https://wavect.io/team/christof-jori/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Christof Jori", "url": "https://wavect.io/team/christof-jori/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } } ], "legalRepresentative": [ { "@id": "https://wavect.io/team/kevin-riedl/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Kevin Riedl", "url": "https://wavect.io/team/kevin-riedl/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } }, { "@id": "https://wavect.io/team/christof-jori/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Christof Jori", "url": "https://wavect.io/team/christof-jori/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } } ], "name": "Wavect GmbH", "subjectOf": { "@id": "https://wavect.io/verified-claims.json#dataset", "@type": "Dataset", "creator": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] }, "description": "A machine-readable registry of quantitative and qualitative claims published by Wavect, with review dates, localized page appearances and public third-party citations where available.", "inLanguage": "en", "isAccessibleForFree": true, "license": "https://creativecommons.org/licenses/by/4.0/", "name": "Wavect verified publication claims", "url": "https://wavect.io/verified-claims.json" }, "url": "https://wavect.io/" }, { "@id": "https://wavect.io/team/kevin-riedl/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Kevin Riedl", "sameAs": [ "https://www.wikidata.org/wiki/Q139796365", "https://www.linkedin.com/in/wsdt", "https://github.com/wsdt" ], "url": "https://wavect.io/team/kevin-riedl/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } }, { "@id": "https://wavect.io/team/christof-jori/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Christof Jori", "sameAs": [ "https://www.wikidata.org/wiki/Q139796367", "https://www.linkedin.com/in/jocr77/", "https://github.com/jo-chris" ], "url": "https://wavect.io/team/christof-jori/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } }, { "@id": "https://wavect.io/#website", "@type": "WebSite", "inLanguage": [ "en", "de", "es", "zh" ], "name": "Wavect", "potentialAction": { "@type": "SearchAction", "query-input": "required name=search_term_string", "target": { "@type": "EntryPoint", "urlTemplate": "https://wavect.io/search/?q={search_term_string}" } }, "publisher": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] }, "url": "https://wavect.io/" }, { "@id": "https://wavect.io/blog/tencent-hy4-preview-coding-agent-review/#webpage", "@type": "WebPage", "dateModified": "2026-09-01", "inLanguage": "en", "isPartOf": { "@id": "https://wavect.io/#website", "@type": "WebSite" }, "lastReviewed": "2026-09-01", "url": "https://wavect.io/blog/tencent-hy4-preview-coding-agent-review/" } ] } ``` ```json { "@context": "https://schema.org", "@type": "BlogPosting", "abstract": "Tencent Hy4 preview is a credible pilot candidate for long-running coding agents, not an automatic replacement for a proven production model. Tencent reports 770B backbone parameters, 49B active per token, a 1M-token context window and an Apache 2.0 release. Its internal 203-task evaluation narrowly beat GLM-5.3 and Kimi K3, while public benchmark records show strong coding results, but those figures do not prove codebase retention or reliability on your stack. Hosted access is the practical starting point because official self-hosting guidance targets 8 B300 or 16 B200 GPUs. Run representative repository tasks, measure accepted changes rather than raw tokens, verify tool use and security, and treat the preview's overthinking and over-verification as measurable cost risks.", "articleBody": " Blog overview/AI and agents/Models and infrastructure Tencent Hy4 Preview Review: Is the 1M-Context Coding Model Worth a Pilot? TL;DR Tencent Hy4 preview is a credible pilot candidate for long-running coding agents, not an automatic replacement for a proven production model. Tencent reports 770B backbone parameters, 49B active per token, a 1M-token context window and an Apache 2.0 release. Its internal 203-task evaluation narrowly beat GLM-5.3 and Kimi K3, while public benchmark records show strong coding results, but those figures do not prove codebase retention or reliability on your stack. Hosted access is the practical starting point because official self-hosting guidance targets 8 B300 or 16 B200 GPUs. Run representative repository tasks, measure accepted changes rather than raw tokens, verify tool use and security, and treat the preview's overthinking and over-verification as measurable cost risks. Yes, Tencent Hy4 preview is worth a controlled coding-agent pilot if your work is limited by long context, multi-step tool use, or repeated repository retrieval. It is not yet a safe default for every production workflow. The model is enormous, self-hosting is a datacenter project, and Tencent explicitly labels this an early version that can reason and verify longer than necessary. Tencent released Hy4 preview on 28 August 2026 with 770 billion backbone parameters, 49 billion active parameters per token, and a context window above one million tokens. The official Tencent announcement also lists WorkBuddy, CodeBuddy, Tencent Cloud TokenHub, and OpenRouter as access routes. Launch access in WorkBuddy and CodeBuddy was announced as free for two weeks. This review was checked on 1 September 2026. It owns one narrow search intent: whether a software team should pilot Hy4 preview for long-horizon coding work. For a market-wide model shortlist, use our separate open-weight LLM comparison. For the architecture choice between retrieval, tuning, and large context, use the RAG vs fine-tuning vs long-context guide. Hy4 preview facts in 60 seconds Tencent Hy4 preview buyer summary, checked 1 September 2026 QuestionVerified answerBuyer implication What is it?A 770B-total, 49B-active Mixture-of-Experts language modelLarge capacity with sparse activation, not a 49B checkpoint How much context?1,048,576 tokens in the serving recipeEnough capacity for very large inputs, but capacity is not proof of recall Is it open?Weights and code under Apache 2.0Commercial use and self-hosting are possible, subject to a normal legal review Can it call tools?Yes, official vLLM and SGLang recipes include reasoning and tool parsersAgent integration is supported, but your harness still owns permissions and recovery What does the official API cost?$0.834 input, $2.501 output, and $0.042 cache hit per 1M tokensLong prompts can be affordable when cache reuse is real Can a team self-host it?Yes, with official recipes targeting 8 B300 or 16 B200 GPUsHosted evaluation should come before infrastructure procurement Main known limitation?Overthinking and over-verification on complex workMeasure time, output volume, and cost per accepted change What is Tencent Hy4 preview? Hy4 preview is Tencent Hy Team's flagship sparse Mixture-of-Experts model for coding, office work, analysis, games, and scientific tasks. The official Hugging Face model card documents 78 backbone layers. The first uses a dense feed-forward network; the other 77 contain 256 routed experts and one shared expert. Each token activates eight routed experts plus the shared expert. The 770B figure covers the backbone. A native multi-token-prediction layer adds another 10B total parameters and 0.7B active parameters for speculative decoding, which explains why some model listings show 780B. The architecture combines Gated DeepSeek Sparse Attention, cross-layer IndexCache reuse, and identity Hyper-Connections. The practical goal is to make long inputs and sustained reasoning more efficient without activating the full model for every token. The weights, an FP8 variant, fine-tuning code, deployment guidance, and license are available in the official Tencent Hy4 preview repository. That is meaningful openness. It does not make a 770B model a laptop download. What did our WorkBuddy coding test show? We asked Hy4 preview inside WorkBuddy to build a fully interactive AI model benchmark dashboard. The useful result was not that it produced HTML. It also wrote the core logic, built headless harnesses, verified JavaScript against a stub DOM, and checked filters plus cost-simulation edge cases before delivery. That is promising evidence of workflow completion, not an independent benchmark. One successful dashboard does not establish a general success rate, security level, or economic advantage. It does reveal a good pilot hypothesis: Hy4 may be valuable when an agent must retain requirements, implementation state, test evidence, and tool output across a longer run. How strong are the Hy4 preview", "articleSection": "AI Models", "author": { "@id": "https://wavect.io/team/kevin-riedl/#person", "@type": "Person", "name": "Kevin Riedl", "sameAs": [ "https://www.wikidata.org/wiki/Q139796365", "https://www.linkedin.com/in/wsdt", "https://github.com/wsdt" ], "url": "https://wavect.io/team/kevin-riedl/" }, "citation": [ { "@type": "WebPage", "name": "official Tencent announcement", "url": "https://www.tencent.com/tencent-releases-and-open-sources-tencent-hy4-preview/" }, { "@type": "WebPage", "name": "official Hugging Face model card", "url": "https://huggingface.co/tencent/Hy4-preview" }, { "@type": "WebPage", "name": "official Tencent Hy4 preview repository", "url": "https://github.com/Tencent-Hunyuan/Hy4-preview" }, { "@type": "WebPage", "name": "LongCodeBench research paper", "url": "https://arxiv.org/abs/2505.07897" }, { "@type": "WebPage", "name": "official vLLM recipe", "url": "https://recipes.vllm.ai/tencent/Hy4-preview" } ], "dateModified": "2026-09-01", "datePublished": "2026-09-01", "description": "Tencent Hy4 preview is a credible pilot candidate for long-running coding agents, not an automatic replacement for a proven production model. Tencent reports 770B backbone parameters, 49B active per token, a 1M-token context window and an Apache 2.0 release. Its internal 203-task evaluation narrowly beat GLM-5.3 and Kimi K3, while public benchmark records show strong coding results, but those figures do not prove codebase retention or reliability on your stack. Hosted access is the practical starting point because official self-hosting guidance targets 8 B300 or 16 B200 GPUs. Run representative repository tasks, measure accepted changes rather than raw tokens, verify tool use and security, and treat the preview's overthinking and over-verification as measurable cost risks.", "headline": "Tencent Hy4 Preview Review: Is the 1M-Context Coding Model Worth a Pilot?", "image": "https://wavect.io/img/blog/headers/header_tencent-hy4-preview-coding-agent-review.svg", "inLanguage": "en", "keywords": "AI Models, Coding Agents", "mainEntityOfPage": { "@id": "https://wavect.io/blog/tencent-hy4-preview-coding-agent-review/", "@type": "WebPage" }, "publisher": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] }, "url": "https://wavect.io/blog/tencent-hy4-preview-coding-agent-review/", "wordCount": 2440 } ``` ```json { "@context": "https://schema.org", "@type": "BreadcrumbList", "itemListElement": [ { "@type": "ListItem", "item": "https://wavect.io/", "name": "Home", "position": 1 }, { "@type": "ListItem", "item": "https://wavect.io/blog/overview/", "name": "Blog overview", "position": 2 }, { "@type": "ListItem", "item": "https://wavect.io/blog/topics/ai-agents/", "name": "AI and agents", "position": 3 }, { "@type": "ListItem", "item": "https://wavect.io/blog/clusters/models-infrastructure/", "name": "Models and infrastructure", "position": 4 }, { "@type": "ListItem", "item": "https://wavect.io/blog/tencent-hy4-preview-coding-agent-review/", "name": "Tencent Hy4 Preview Review: 1M Context & Cost | ", "position": 5 } ] } ``` ```json { "@context": "https://schema.org", "@type": "FAQPage", "mainEntity": [ { "@type": "Question", "acceptedAnswer": { "@type": "Answer", "text": "Hy4 preview is a Tencent Hy Team Mixture-of-Experts language model aimed at coding, agent workflows, office work, analysis, games, and science. Its backbone has 770 billion total parameters, with 49 billion activated per token, plus a separate 10-billion-parameter MTP layer for speculative decoding." }, "name": "What is Tencent Hy4 preview?" }, { "@type": "Question", "acceptedAnswer": { "@type": "Answer", "text": "Yes. The model card lists a 1M context length and the vLLM recipe specifies 1,048,576 tokens. That is capacity, not proof that the model recalls and uses every code detail accurately at maximum length. Test retrieval and task success across several context bands." }, "name": "Does Hy4 preview really support one million tokens?" }, { "@type": "Question", "acceptedAnswer": { "@type": "Answer", "text": "Tencent publishes the model weights, FP8 weights, code, fine-tuning guidance, and deployment recipes under Apache 2.0. Teams should still review the license, third-party components, model behavior, and their own regulatory duties before production use." }, "name": "Is Hy4 preview open source?" }, { "@type": "Question", "acceptedAnswer": { "@type": "Answer", "text": "Tencent's 28 August announcement lists 0.834 US dollars per million input tokens, 2.501 dollars per million output tokens, and 0.042 dollars per million cache-hit tokens. Confirm the live price and conditions with the provider route you plan to use." }, "name": "How much does the Hy4 preview API cost?" }, { "@type": "Question", "acceptedAnswer": { "@type": "Answer", "text": "Not the official full deployment. The vLLM recipe lists 16 B200 or 8 B300 GPUs for the FP8 model. Community quantizations may reduce memory needs, but they require separate quality, performance, and license verification." }, "name": "Can Hy4 preview run on a laptop or one GPU?" }, { "@type": "Question", "acceptedAnswer": { "@type": "Answer", "text": "Tencent's blind internal evaluation put Hy4 slightly ahead on its 203 engineering tasks. The margins were small and the evaluation was vendor-run. Choose from a controlled test on your repositories, tools, languages, limits, and acceptance criteria." }, "name": "Is Hy4 preview better than GLM-5.3 or Kimi K3?" }, { "@type": "Question", "acceptedAnswer": { "@type": "Answer", "text": "The largest practical risks are treating context capacity as reliable memory, paying for unnecessary reasoning and verification, trusting tool calls without external controls, and choosing datacenter-scale self-hosting before a hosted pilot proves business value." }, "name": "What is the biggest Hy4 preview risk for coding agents?" } ] } ```