Tencent Hunyuan Hy4 Officially Released, Benchmarked

Episode 1 · Tencent Hunyuan Launches Flagship Hy4 with Vision and Computer Use (2026-11-04, 29 posts)

On November 4, Tencent Hunyuan officially released its flagship model Hunyuan Hy4. Compared with the preview version, the official release significantly scales up both pretraining and post-training, adds vision capabilities and computer-use support for the first time, and is claimed by the company to rank among the strongest open-source models overall.

Confirmed

• Hy4 adds fine-grained image processing, chart analysis, and command-line vision and computer-operation capabilities; it can drive terminals and control a computer via the screen, making it the first vision-capable model in the Hy series
• The model is now available on Tencent Cloud, OpenRouter, CodeBuddy, WorkBuddy, and Hy AI Studio
• The company announced it will open-source the Hy4 weights
• It scores 78.9 on the OSWorld 2.1 benchmark, narrowing the gap with Claude Opus 5.5 (81.8) to single digits, ahead of GLM 5.3, DeepSeek-V4.1-Flash, Kimi K3.1 Pro, and GPT 6.1
• Official use cases listed include finance, frontend development, gaming, office work, and research, with emphasis on responsible building in cybersecurity and biosafety
• Blogger XianbaoQIAN, resharing the release news, praised the HY series for its honest evaluation practices and noted that paper author Shunyu Yao contributed to the work; the shared chart covers a 12-benchmark comparison, with Hy4 standing out on ProgramBenCh, FrontierCode 2, and others

Why it matters

Hy4 is closing in on frontier closed-source models on benchmarks like OSWorld that reflect real-world computer use; combined with the plan to open-source the weights, it further narrows the open/closed gap on agent and computer-use tasks, giving the open-source community a strong new contender.

• Tencent Hunyuan launches Hy4: vision, computer use, and top open-source benchmark results — 腾讯混元 · 2026-11-04
• HY4 Shows Huge Capability Jump Over Preview, Wins Praise for Honest Evals — Xianbao_QIAN · 2026-11-04
• Tencent launches flagship Hy4 with native vision, plans open weights at same API price — 智东西 · 2026-11-04
• Tencent Hunyuan Launches Hy4 with Vision, Claims Top Tier Among Open Models — TencentHunyuan · 2026-11-04
• Tencent Hunyuan Hy4 launches: top open-source model gains vision and long-horizon agent skills — APPSO · 2026-11-04
• Tencent Hunyuan unveils Hy4, open weights coming soon — heatOverflower · 2026-11-04
• Tencent launches flagship Hy4 model, open weights and system card promised soon — testingcatalog · 2026-11-04
• Tencent Hy4 benchmark charts: 78.9 on OSWorld, closing on Claude Opus 5.5 — heyshrutimishra · 2026-11-04
• Tencent's Hy4 scores 78.9 on OSWorld and the weights will be open-sourced — heyshrutimishra · 2026-11-04
• Hy4 tops open models in finance eval at 3.04, hallucination rate cut to 5.45% — heyshrutimishra · 2026-11-04
• Tencent's Hy4 rebuilds 7M-cell embryo in 1.5h, trains self-improving models — heyshrutimishra · 2026-11-04
• Tencent's Hy4 agent self-trained a smaller model that beat Tencent's own baseline — heyshrutimishra · 2026-11-04
• Tencent Hunyuan ships Hy4 with vision, touting top-tier open-model status — mishig25 · 2026-11-04
• Tencent's Hunyuan Hy4 adds vision and computer use, but long-task reliability remains untested — immajusteeyy · 2026-11-04
• Tencent Hunyuan launches Hy4 with vision and computer use, claims top open-model tier — Intelligent_Ad2105 · 2026-11-04
• Hunyuan Hy4 hands-on: 770B MoE with native vision builds 3D keyboard, theme park and more — 量子位 · 2026-11-04
• Tencent's Hy4 adds vision and computer use — can it survive long real-world tasks? — Short_Association904 · 2026-11-04
• Tencent touts Hy4 as top open model — but weights aren't out and rivals' scores are its own runs — mawcopolow · 2026-11-04
• Tencent Hunyuan Hy4 official release: first native vision, OSWorld 2.1 at 78.9 — AGI Hunt · 2026-11-04
• Tencent Hunyuan launches Hy4 with vision, claiming top-tier open-weight performance — cihangxie · 2026-11-04

9 more related posts →

Episode 2 · Tencent Hunyuan Hy4 scores 35 on Intelligence Index, second among Chinese major models (2026-11-05, 2 posts)

Artificial Analysis evaluated Tencent's latest Hunyuan Hy4, which scored 35 on the Intelligence Index v5, tying with Grok 4.7 (xhigh) and ranking second among Chinese major-vendor models behind Kimi K3.1 Pro, though it ranked low on output verbosity.

• Tencent Hy4 scores 35 on AA Intelligence Index v5, the #2 model from a Chinese AI lab — ArtificialAnlys · 2026-11-05
• Tencent Hy4 scores 35 on Artificial Analysis Intelligence Index, ranks #43 of 106 — ArtificialAnlys · 2026-11-05