Tencent Hy3 / Hy4 preview (formerly Hunyuan)
A Tier · 8.1/10
Tencent Hy (formerly Hunyuan) open-weight MoE family -- Hy4 preview (2026-08-28) is the new flagship at 770B total and 49B active with a 1M-token context, Apache 2.0, at $0.834 in / $2.501 out per 1M tokens on Tencent Cloud TokenHub. Hy3 (GA 2026-07-06; 295B total / 21B active, 256K context, Apache 2.0) remains the budget pick at $0.132 / $0.528. Integrated into Yuanbao, WorkBuddy, CodeBuddy, and ima
Score Breakdown
Personality & Tone
Tencent's first serious open-weights LLM
Tone: Direct and structured, similar to DeepSeek's chat tone -- compact answers in technical domains, reasonable Chinese-language polish, English noticeably less expressive than Claude or GPT.
Quirks: PRC content filters apply across the same set of regulated topics as DeepSeek and Qwen. Distribution via Yuanbao, WorkBuddy, CodeBuddy, and Weixin means many Hy users in the wild are talking to it through Tencent app surfaces rather than a developer console.
The Good and the Bad
What we like
- +Open weights from a top-3 Chinese tech company is itself the headline -- Tencent had been a notable holdout while DeepSeek, Alibaba, Moonshot, MiniMax, and Z.ai shipped open-weight flagships through 2024-25
- +Pricing is aggressive -- Hy3 lists at $0.132 in / $0.528 out per 1M tokens on Tencent Cloud TokenHub, and even the much larger Hy4 preview is $0.834 / $2.501
- +Context is comfortable for retrieval-heavy and agentic workflows -- 256K on Hy3 and a 1M-token window on Hy4 preview
- +Distribution moat: Tencent already runs Hy3 across Yuanbao, WorkBuddy / CodeBuddy, ima, Marvis, Weixin Official Account assistants, and games. The model meets users where they already are, which matters in the Chinese market more than benchmarks
- +The family is moving fast -- the April preview, Hy3 GA on 2026-07-06, then Hy4 preview on 2026-08-28 -- and Tencent says the next batch of Hy4-series models is expected to roll out soon
What could be better
- −Much of the headline evidence is Tencent-run: Hy3's 270-expert blind evaluation (2.67/4 vs GLM-5.1 at 2.51/4) and Hy4 preview's 163-expert evaluation (2.99/4 vs GLM-5.3 at 2.92 and Kimi K3 at 2.94) are internal -- treat them with the usual self-reporting discount until independent runs corroborate
- −PRC content filters apply -- expect refusals or blandness on Tiananmen, Taiwan sovereignty, Xi Jinping, and other politically sensitive topics. Pattern matches DeepSeek and Qwen
- −Geo-availability improved: both current releases (Hy3 and Hy4 preview) ship under Apache 2.0, so the preview-era community license's EU / UK / South Korea exclusion no longer applies to current weights. Developers should still check local data-residency rules before deploying via API; self-hosted open-weights deployment remains the fallback
- −English-language polish lags Claude and GPT for nuanced writing tasks -- this is a coding / reasoning / Chinese-language pick first, prose-quality pick second
- −Hy4 preview is explicitly early -- Tencent's model card lists known issues including spending longer than necessary reasoning through complex tasks and a tendency to over-verify its own work, and 770B / 49B-active weights need datacenter-class multi-GPU serving
Pricing
Free (consumer)
- ✓Yuanbao app (Tencent's consumer chatbot), including its free Agent function
- ✓WorkBuddy and CodeBuddy (Hy4 preview was free there for two weeks after launch)
- ✓ima and other Tencent products
- ✓Basic usage limits apply
API (Hy4 preview)
- ✓$2.501 per 1M output tokens, $0.042 per 1M cache-hit tokens (Tencent Cloud TokenHub, international pricing updated 2026-11-01)
- ✓770B total / 49B active MoE, 1M-token context
- ✓Also on OpenRouter (tencent/hy4-preview, $0.7506 in / $2.2509 out list price)
API (Hy3)
- ✓$0.528 per 1M output tokens, $0.033 per 1M cache-hit tokens (Tencent Cloud TokenHub, international pricing)
- ✓295B total / 21B active MoE, 256K context
- ✓OpenRouter lists tencent/hy3 at $0.0825 in / $0.33 out
Self-hosted (open-source)
- ✓Open weights on Hugging Face and ModelScope: tencent/Hy3 and tencent/Hy4-preview (plus FP8 builds)
- ✓Apache 2.0 for both Hy3 and Hy4 preview
- ✓Hy3: Tencent recommends 8 GPUs such as H20-3e or larger-memory GPUs; Hy4 preview reference commands use 8-way tensor parallel on the FP8 build
System Requirements
Hardware needed to self-host. Min = smallest viable setup (usually heavy quantization). Max = full-precision / production-grade.
| Model variant | Min | Max |
|---|---|---|
| Hy3 (295B total, 21B active MoE)Apache 2.0. Weights: tencent/Hy3 and tencent/Hy3-FP8 on Hugging Face. 21B-active means inference is cheaper than the 295B total suggests, but production serving wants datacenter GPUs | 8 GPUs -- Tencent recommends H20-3e or other GPUs with larger memory | 8-GPU datacenter node (BF16; Hy3-FP8 build also published) |
| Hy4 preview (770B total, 49B active MoE)Apache 2.0. Weights: tencent/Hy4-preview and tencent/Hy4-preview-FP8. Built-in MTP layer for speculative decoding | 8 GPUs with tensor parallel 8 on the FP8 build (Tencent vLLM / SGLang reference commands) | Multi-GPU datacenter node for the 1M-token context |
Known Issues
- HY4 PREVIEW RELEASED (2026-08-28): Tencent released and open-sourced Hy4 preview -- 770B total / 49B active MoE with a context window exceeding 1M tokens, Apache 2.0 weights on Hugging Face (tencent/Hy4-preview, plus Hy4-preview-FP8) and ModelScope. API via Tencent Cloud TokenHub at USD 0.834 per 1M input tokens, 2.501 per 1M output and 0.042 for cache hits, and via OpenRouter. Tencent's model card flags known issues (over-long reasoning, over-verification) and Tencent says the next batch of Hy4-series models is expected soon. Hy3 remains the cheaper, stable optionSource: Tencent: Tencent Releases and Open-Sources Tencent Hy4 preview (2026-08-28); huggingface.co/tencent/Hy4-preview · 2026-08-28
- Naming and official site: Tencent now refers to 'Tencent Hunyuan (now known as Tencent Hy)'. The official site per Tencent's Hugging Face model cards is aistudio.tencent.com (Hy AI Studio). hy3ai.com, previously listed on this page as the product site and as a web-access route, is a third-party community information page about Hy3 Preview, not a Tencent propertySource: Tencent Hy3 announcement (tencent.com), huggingface.co/tencent/Hy3, hy3ai.com footer · 2026-11-02
- HY3 REACHED GA (2026-07-06): Tencent officially released the final Hy3 model, upgrading the preview launched on 23 April with strengthened reinforcement learning and higher-quality data. Same architecture (295B total / 21B active MoE, 256K context) but two material changes: (1) it ships under **Apache 2.0** with the preview's **EU / UK / South Korea use restriction lifted**, open-weights day-one on Hugging Face + ModelScope; (2) heavily strengthened autonomous-agent capability -- Caixin reported a **~90% task-completion rate** across several Tencent internal applications, and Yuanbao added a free Agent function. API is served via **Tencent Cloud's TokenHub** platformSource: Tencent (tencent.com/en-us/articles/2202386.html), Caixin Global (2026-07-06), TechNode (2026-07-07), Pandaily · 2026-07-06
- Current list pricing (Tencent Cloud International model pricing, page updated 2026-11-01, USD per 1M tokens): Hy3 0.132 input / 0.528 output / 0.033 cache; Hy4 preview 0.834 / 2.501 / 0.042. Hy3 Preview no longer appears in that price list. Earlier versions of this page quoted ~1.2 RMB per 1M input tokens for Hy3 Preview; mainland CNY pricing was not re-checkedSource: Tencent Cloud: Model pricing (tencentcloud.com/document/product/1300/78937) · 2026-11-01
- PRC content filtering applies. Hy3 declines to engage on Tiananmen Square, Taiwan sovereignty, and other regulated topics. Same operating reality as DeepSeek, Qwen, Kimi, GLM -- if your use case needs unfiltered geopolitics, choose a Western frontier model or a community-tuned open-weights variantSource: Pattern from DeepSeek/Qwen; Hy3 follows the same regulatory regime · 2026-04
Best for
Chinese-market builders, multilingual products that need strong CN performance, and cost-sensitive developers who want a cheap open-weight MoE API (Hy3 at $0.132 per 1M input tokens) or a 1M-context open flagship (Hy4 preview). Also anyone watching the open-weights field who wants Tencent's distribution-backed entry on the bench.
Not for
English-first creative writing (Claude or GPT are stronger), use cases that touch Chinese-government-sensitive topics, or teams that need a fully proven track record of third-party benchmark verification. Hy4 preview is explicitly preview-stage; Hy3 is the stable release.
Our Verdict
Tencent Hy has gone from open-weights holdout to a fast-shipping lab. Hy3 (GA July 2026, 295B / 21B active, Apache 2.0) is a cheap, stable production option at $0.132 / $0.528 per 1M tokens, and Hy4 preview (August 2026, 770B / 49B active, 1M context) is, by Tencent's account, among the top tier of open-source models at $0.834 / $2.501. Distribution through Yuanbao, WorkBuddy, CodeBuddy, and Weixin is a structural advantage Western developers tend to under-weight. Caveats: the strongest evidence is still Tencent-run, Hy4 is a preview with known over-reasoning issues, and PRC content filters apply. Worth a spot on the bench for any team comparison-shopping cheap open-weight MoE APIs, especially if Chinese market reach matters.
Sources
- Tencent: Tencent Releases and Open-Sources Tencent Hy4 preview (2026-08-28) (accessed 2026-11-02)
- Tencent: Hunyuan officially releases Hy3 (GA, 2026-07-06) (accessed 2026-11-02)
- Hugging Face: tencent/Hy4-preview model card (accessed 2026-11-02)
- Hugging Face: tencent/Hy3 model card (accessed 2026-11-02)
- Tencent Cloud: Model pricing (TokenHub, international) (accessed 2026-11-02)
- OpenRouter models API (tencent/hy3, tencent/hy4-preview) (accessed 2026-11-02)
- Hy AI Studio (Tencent official site) (accessed 2026-11-02)
- Caixin Global: Tencent launches final Hunyuan 3 model with free AI-agent feature (2026-07-06) (accessed 2026-07-22)
- TechNode: Tencent launches Hunyuan Hy3, integrates across products (2026-07-07) (accessed 2026-07-22)
- Caixin Global: Tencent unveils new AI model to close gap with rivals (2026-04-23) (accessed 2026-04-25)
- HuggingFace: tencent/Hy3-preview (accessed 2026-04-25)
- Pandaily: How Tencent is building its global AI moat (accessed 2026-04-25)
- Dataconomy: Tencent uses product rollout to define Hy3 Preview (accessed 2026-04-25)
- hy3ai.com (third-party community information page, not Tencent) (accessed 2026-11-02)
Explore more Tencent Hy3 / Hy4 preview (formerly Hunyuan) rankings
Deeper leaderboards, benchmarks, task-specific tier lists, and status/pricing pages for Tencent Hy3 / Hy4 preview (formerly Hunyuan).
The Tier List Tuesday
Weekly newsletter: tier movers, new entrants, and the VS of the week. Built from our daily AI-tool sweeps. No spam, unsubscribe anytime.
Alternatives to Tencent Hy3 / Hy4 preview (formerly Hunyuan)
Claude (Anthropic)
**Claude Duet opened as a research preview on 2026-10-21: a proactive agent app on iOS, Android and claude.ai, built around Claude Opus 5.5, that sends a morning brief from your connected apps, keeps working on a cloud computer while you are away and asks before it sends, pays or deletes anything. It is included in Pro and Max in the US with a waitlist, on the same usage limits as the rest of Claude.** **Oct 23: chat and Cowork became one Claude for Enterprise (GA). Oct 22: the full Claude Platform went GA on Google Cloud as an Anthropic-managed offering billed through Google Cloud. Oct 26: Anthropic filed an S-1 for an IPO on Nasdaq under ANT.** **Claude Haiku 5.5 shipped on 2026-10-07 at $0.10 / $0.50 per 1M tokens for prompts up to 100K ($0.50 / $2.50 above) -- 90% below Haiku 4.5 on most requests -- with adjustable effort for the first time on a Haiku, and the same day Anthropic halved Sonnet 5.5 cache reads to $0.10 and added monthly API credits for Max ($100 or $200) and Team (up to $500 pooled).** **Anthropic expanded the Cyber Verification Program on 2026-10-06 into three access tiers (Defense, Red Team, Specialized) that unlock reduced cyber safeguards on Opus 5.5, Sonnet 5.5, Mythos 5.1 and future models for verified security teams -- closing the 'coming weeks' note Opus 5.5 shipped with; generally available Claude keeps its conservative cyber blocks.** **Two Claude 5.5 models in six days: Claude Opus 5.5 (2026-09-22) and Claude Sonnet 5.5 (2026-09-28).** Opus 5.5 performs at Fable 5.1 level on most work and costs 40% less than Opus 5 to run -- $4/$20 per 1M (down from $5/$25) with cache reads cut to $0.20 (from $0.50), fast mode at $8/$40 up to 2.5x speed, output 30%+ faster, and five-hour usage limits raised on Pro, Max, Team and seat-based Enterprise with a saveable rate-limit reset. Sonnet 5.5 keeps Sonnet 5's $2/$10 but needs far fewer tokens (up to 30% cheaper per task), runs 30%+ faster and jumps Terminal-Bench 4.0 from 10.3% to 70.6%. Both are 1M context, Jun 2026 cutoff, on the Claude Platform, Bedrock, Vertex and Foundry; Haiku 5.5 followed on 2026-10-07. Opus 5.5 launches with Fable-class cyber and bio safeguards (most cyber tasks fall back to Opus 4.8) and cannot run with thinking off.
Claude Mythos 5.1
**From 2026-10-06 Mythos 5.1 is included in every tier of Anthropic's expanded Cyber Verification Program, and existing Project Glasswing members move into its Specialized Access tier without reapproval -- with the first disclosed Glasswing numbers: at least 129,000 verified vulnerabilities found April-July 2026, more than 33,000 rated critical or high.** Anthropic's trusted-access frontier model. **Mythos 5.1 launched 2026-09-01** alongside Fable 5.1, and Anthropic now states outright that they are **the same model with different safeguards** -- the 5.1-cycle gap is 60.9% vs 55.8% on Terminal-Bench 4.0, which Anthropic attributes to safeguard interventions rather than capability and expects to shrink. Originally launched June 9, 2026 alongside Claude Fable 5. Suspended June 12 by a US export-control order, then PARTIALLY RESTORED July 1, 2026 (US government lifted controls June 30): Mythos 5 is back for a set of US organizations with government approval, while Anthropic works to re-expand the broader Glasswing program. Public Fable 5 returned globally the same day. Gated to Project Glasswing orgs + select biology researchers.
Gemini (Google)
**Oct 29: the Gemini app can now generate, preview and edit documents, PDFs, slide decks, spreadsheets, Markdown and code right inside the chat for all users (PDF on web and mobile, the other formats on web only). Oct 30: the Computer Use tool went generally available in the Gemini API across the Interactions and GenerateContent APIs.** **Gemini 4 Carbon was announced on 2026-10-28 as Google's strongest model, but it is not generally available yet: access starts this week for a subset of cyber defenders through the Fairwind Program, and Google says developers, businesses and consumers get it later. When it opens it carries Argon's intro price of $2 / $10 per 1M tokens, rising to $4 / $20 after the intro period. Google reports it #1 on the Artificial Analysis Intelligence Index v5.0 at 59.** **Gemini 4 Argon went generally available on 2026-10-13: in the Gemini API (paid tier only), AI Studio, Gemini for business and GitHub Copilot, in the Gemini app and AI Mode in Search for Google AI Pro and Ultra subscribers, and in Google Antigravity for Ultra. The API rate card prices it at $2 input / $10 output per 1M tokens with $0.10 implicit-cache reads through November 30, 2026, doubling to $4 / $20 / $0.20 on December 1; there is no free tier and no explicit caching.** Google calls it its flagship reasoning model, with a 1M-token unified input-output limit, #1 on the Vals Index, 87% on the ERQA embodied-reasoning benchmark and 0.7% attack success on Gray Swan prompt injection, alongside the 9/30 numbers (77.9% DeepSWE v1.1, 51.3% AutomationBench, 91.7% LVBench, 68% CWE-bench v1) -- all vendor-reported. Earlier in October: Gemini 3.7 Flash and 3.5 Flash were retired from AI Studio (10/08) with API calls auto-routed to 3.8 Flash and 3.6 Flash, Nano Banana 2.1 went GA (10/06) and SynthID Detector opened to everyone in English (10/07). Late September in the Gemini app: **skills replace Gems** (9/30; slash-command instructions with reference files, Gems removed from November 2026 for personal accounts), **Guided Vision in Gemini Live** for blind and low-vision users on Android (10/01), and the 9/23-9/24 wave still stands: Gemini 3.8 Flash TTS and Flash-Lite TTS with voice design at $0.50/$9.00 and $0.50/$6.00 per 1M (doubling 2027-01-01), Gemini 3.8 Live with Live Avatar in Gemini Enterprise, free 1080p Omni 1.1 video in Google Vids. Googlebook was scheduled to reach shelves 10/04 (shipment not yet vendor-confirmed).
Grok
**Nov 2: the grok-imagine-image-quality API slug is retired; requests to it, and to grok-imagine-image-pro, which already redirected there, are now answered by grok-imagine-image-2.0 at low quality, which xAI prices $0.01 per image below the old model at every resolution ($0.04 / $0.05 / $0.06 at 1K / 1.5K / 2K).** **The X Subscription (2026-10-12) folds Grok, Grok Bot, Cursor and X Premium into one plan: Free, Plus $8, Premium $30 (what SuperGrok, Cursor Pro and X Premium+ cost $90 for combined), Super $100 and Ultra $200 a month, with one weekly-refreshing usage pool shared across Grok and Cursor; the old SuperGrok, Cursor and X Premium plans are closed to new sign-ups.** **Team Bots launched 2026-09-28 in public beta on Teams and Enterprise plans** -- shared Grok Bots built around a role, with team context (files, instructions, skills), plugins (Salesforce, Notion, GitHub), credentials for APIs without a plugin, and memories, where the bot is shared but each person's conversations and memories stay private; each Team Bot gets its own Slack handle. SpaceXAI runs one per major sales account, and its 9/22 case study says the merged SpaceXAI + Cursor support desk absorbed a 175% ticket increase on Grok Bot with no new hires at $0.20-$0.30 per resolution. **Grok 4.7 (9/21)** remains the flagship at $2/$6 per 1M with chart-only benchmarks; Grok Voice Transcribe 2.0 and the corrected $15/1M-character TTS price stand.
Muse Spark + Muse Wonder (Meta)
Meta Superintelligence Labs' model family, now four lines deep: **Muse Wonder** (2026-10-12, frontier, $4/$20 per 1M on the Meta Model API; Wonder Full only in Muse Code subscriptions), **Muse Spark 1.3** (2026-09-02, $1.25/$4.25, or $0.10/$0.20 in Contributor mode if Meta may train on your content), **Muse Spark Open** (2026-10-15, 816B/42B-active open weights, 1M context, runs on two Mac Studios with M5 Ultra) and **Muse Glimmer** (2026-08-10, 30B, Apache 2.0, built for local agents). Consumers get it through **Muse**, Meta's personal-agent app (2026-09-08): free with limits, Power $20/mo, Maximum $100/mo
GPT-Rosalind (OpenAI)
**Since 2026-10-15 US university labs can apply for the Rosalind Academic Plan: up to 30 ChatGPT Business Standard seats free for up to 12 months, with GPT-Rosalind and Rosalind Workbench behind trusted-access approval.** OpenAI's first domain-specific model -- life sciences, drug discovery, translational medicine. Launched 2026-04-16 as a Trusted Access research preview. Launch partners: Amgen, Moderna, Allen Institute, Thermo Fisher. Paired with a Life Sciences Codex plugin (50+ scientific tool integrations)
GPT-5.6-Cyber / GPT-5.4-Cyber (OpenAI)
**GPT-5.4-Cyber reached its API removal date on 2026-10-01: the gpt-5.4-cyber model page now returns Not found while gpt-5.6-cyber stays live (checked 2026-10-02; OpenAI's deprecations doc had not yet moved the row to Past Deprecations). Its replacement is `gpt-5.6-cyber`**, an alias for OpenAI's most advanced purpose-trained cyber models, gated behind the Daybreak program and priced at $12.50/$75 per 1M (first published price for any OpenAI cyber model). Original page: OpenAI's defensive-cybersecurity variant of GPT-5.4, launched 2026-04-16. Lowered refusal boundary for security-research tasks and native binary reverse-engineering. Access gated via Trusted Access for Cyber (TAC) program -- thousands of verified defenders, hundreds of teams, no public pricing. On **2026-08-17 OpenAI published its first dedicated post on the Hugging Face incident**, conceding it 'underestimated the real-world cyber capabilities of our AI models' and confirming it now releases cyber capabilities only to trusted defenders. **On 2026-09-01 OpenAI confirmed GPT-6 Astra meets the Critical cyber threshold -- the first model it has ever designated at that level -- and on 2026-09-03 committed $1B to Daybreak for Frontline Defenders**
Microsoft MAI-Thinking-1
Microsoft's first in-house reasoning model -- launched 2026-06-02 at Build as the flagship of seven new MAI models. 35B-active / ~1T-total sparse Mixture-of-Experts, 256K context. AIME 2025 97.0%, matches leading models on SWE-Bench Pro, and beat Claude Sonnet 4.6 in human-preference testing. Available on Microsoft Foundry + OpenRouter / Fireworks / Baseten
MiMo (Xiaomi)
Xiaomi's MiMo family. MiMo-V2.6 (2026-09-22) open-sources Pro (1.02T total / 42B active MoE) and Flash (309B / 15B active) under MIT, both natively omnimodal (text, image, video, audio) with 1M context. API prices unchanged from V2.5: Pro $0.435 input / $0.87 output per 1M tokens, Flash $0.14 / $0.28. MiMo-V2.5 TTS and open-source ASR round out the voice stack