Tencent releases Hy4 Preview, a 770B-parameter open model with a 1M context window, and says it narrowly beat GLM-5.3 and Kimi K3 in internal engineering tests
Tencent’s Hunyuan releases have moved quickly from the 295B-parameter Hy3 preview in April to a July Hy3 release positioned against Z.AI’s GLM-5.1 and GLM-5.2. Hy4 is a much larger escalation in that same open-model campaign, adding a 1M-token context window and a direct internal-test claim against Z.AI and Moonshot.
Z.AI had already made its 754B-parameter GLM-5.1 available under an MIT license, giving Tencent’s earlier GLM-5.1 rival a clear open-model benchmark. The story matters because Tencent is contesting that position with a similarly scaled model rather than differentiating chiefly on fast responses, as it did with Hunyuan Turbo S.
First-order effects
Tencent gives developers an open alternative at the frontier-model scale, centered on engineering work and long-context inputs.
Z.AI and Moonshot face a fresh public comparison point after Tencent’s claim that Hy4 narrowly surpassed GLM-5.3 and Kimi K3 in its internal engineering tests.
Second-order effects
The comparison raises the importance of independent coding, agentic, and long-context evaluations for buyers choosing among Tencent, Z.AI, and Moonshot models rather than relying on vendor benchmarks alone.
Tencent’s larger release sharpens competition among Chinese open-model providers on both model scale and usable context capacity, not solely on licensing terms.
Third-order effects
If successive releases keep compressing the gap between leading open models, frontier capability will increasingly be distributed through weights and deployment ecosystems rather than confined to proprietary APIs.
Long context is becoming a product-level competitive dimension alongside benchmark scores, shifting evaluation toward whether models can handle larger working sets in practical engineering workflows.
The trend:Chinese AI labs are turning open releases into a rapid frontier-model race, pairing larger architectures with long-context capability and benchmark-led positioning.
Tencent just open-sourced a model that outscores GPT-5.6 Sol on coding. It also beats Claude Fable 5 and Grok-4.6. Hy4 preview. 770B parameters. 49B active. 1M context. Weights live on Hugging Face right now. On the Arena AI Code leaderboard it's ranked #5 globally. Only Claude
Tencent has released and open-sourced Hy4 preview, a next-generation large language model with 770B total parameters and a 1M+ token context window. Ranked among the top tier of open-source models. Try Hy4 preview. OpenRouter: https://openrouter.ai/... Tencent Cloud TokenHub:
Big news: Hy4 preview by @TencentHunyuan just landed ~#5 in the Code Arena: WebDev with 1633 pts (AutoEval). This is a significant improvement from Hy3 at #31 overall (+115 pts)! Among open models, Hy4 preview is ~#3, compared to Hy3 at #7. Note: this is an early AutoEval
@TencentHunyuan Shipping it live in @CommandCodeAI 🐐 Excited to be an early partner to benchmark Hy4 preview, in our internal bench it performed better than GLM 5.3 Flash on token costs and time. Will continue to test and share more results.
Hy4 Preview is out! 770B-A49B, 1M context, Apache 2.0 Reported scores are great, putting it roughly on par with GLM-5.3 (!). Tencent has REALLY picked up model training since @ShunyuYao12 joined, congrats!
Introducing Hy4 preview ◆ Open weights: 770B MoE, 49B active, 1M context ◆ Built for real work: code, docs and analysis, scientific research ◆ Hy4 preview helped optimize its own training pipeline, kernels included ◆ We ship upgrades, not price hikes Free in @WorkBuddy_AI
🚀 Hy4 preview is here. 770B, 49B active, 1M context. Built for productivity. Open source frontier. Consistent affordable price. Use it. Tell us what breaks. More on Hy blog： https://hy.tencent.ai/... HuggingFace： https://huggingface.co/... Github：https://github.com/ ...
Chinese AI models are advancing relentlessly! Tencent's Hy4 preview is another near-frontier open-source model. It seems at the Kimi K3 level while being 3 times smaller at 770B parameters! In particular, I am interested in the high scientific reasoning benchmarks! Looking good!
Tencent says it's a early version, and that they will continue to polish it post release, but it already appears to be very strong. They have been increasing their AI spend, looks like it's starting to pay off. Input: $0.834/m Output: $2.501/m