AnswersPublished23 sources
Tencent Expands Hy4 Capacity 10× After WorkBuddy Users Hit Limits
Tencent said it expanded Hy4 inference capacity tenfold and raised WorkBuddy’s free quota after users hit first day limits. Hy4 scored 35 on Artificial Analysis’s Intelligence Index.
Published byEdited with GPT-6 LunaImages generated with GPT Image 2
Research answer
Key takeaways
- Tencent said it expanded Hy4 inference capacity tenfold and raised WorkBuddy’s free quota after users hit first day limits.
- Hy4 scored 35 on Artificial Analysis’s Intelligence Index. Agent Arena’s initial announcement placed it 14th overall, while a later report lists 13th.
AI Prompt
openai.comCreate a landscape editorial hero image for this Studio Global article: What did Tencent announce on November 5, 2026, about its tenfold expansion of Hy4’s inference cluster following a WorkBuddy demand surge and. Article summary: Tencent said on November 5, 2026, that it had expanded Hy4’s inference capacity tenfold after surging WorkBuddy demand caused many users to exhaust their quotas on the first day, and it increased free usage allowances.[7. Topic tags: general, general web, user generated. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts with fa
Tencent said on November 5, 2026, that it had expanded Hy4’s inference capacity tenfold after heavy demand on WorkBuddy drove many users to their usage limits on the first day. The company also increased free usage allowances and said it would keep adjusting resources as demand changed.3
4
WorkBuddy capacity: what changed, and what limits may remain
Hy4 became available in WorkBuddy on November 4, the day Tencent released the full model. After many users hit their quotas, Tencent expanded the inference cluster tenfold and raised the free-use quota.3
4
9
The expansion did not necessarily mean unlimited access. Reporting said high-end computing capacity remained constrained and demand was concentrated at peak times, so temporary limits could still apply.9 For users needing another option, a report cited two workarounds: switch to Hy3 or move non-urgent tasks to the evening. It also said free Hy3 access in WorkBuddy was extended through 23:59 on November 15.
39 That extension concerns Hy3; it is distinct from Tencent’s increase to Hy4’s free-use allowance.
Artificial Analysis: a score of 35
Hy4 scored 35 on Artificial Analysis’s Intelligence Index. That put it second among models from Chinese AI labs, behind Kimi K3.1 Pro at 46. Its score matched Grok 4.7 at xhigh effort and was one point below Claude Haiku 5.5.21
29
A report describing the November 5 index says it covered 724 models and evaluated areas including complex reasoning, coding, tool use, multimodal understanding, long-horizon planning, and scientific problem-solving.39 The supplied material supports Hy4’s placement behind Kimi, but does not establish a complete ranking comparison against Qwen3.8 Max and DeepSeek V4.1 Flash.
Cost comparisons need a caveat. Pandaily reported that Tencent cited an Artificial Analysis figure of about $0.78 per task, which Tencent compared with roughly one-tenth of Grok 4.7’s cost and one-thirteenth of Qwen3.8 Max’s at similar score levels.15 Artificial Analysis’s Hy4 model page lists $1.03 per Intelligence Index task.
38 The supplied excerpts do not explain the difference, so these figures should not be treated as interchangeable.
Agent Arena: strong placement, but reports differ
Agent Arena’s announcement placed Hy4 14th overall, based on more than 8,500 real-world, long-horizon agent sessions. It said Hy4 was still proprietary, but would tie Kimi K3.1 Pro for first among open models once its weights were released. The announcement also placed Hy4 on the score-versus-cost Pareto frontier.28
There is a discrepancy in later coverage: TechNode reported Hy4 at 13th overall and first among open-weight models.3 The supplied reports do not clarify whether that difference reflects a changed leaderboard snapshot or another distinction, so the announcement’s 14th-place result and TechNode’s later 13th-place account should not be collapsed into one ranking.
Pandaily also described Hy4 as being on the Pareto frontier relative to more expensive models such as Claude Opus 5.5.15 A frontier position indicates a favorable score-and-cost trade-off; it does not mean Hy4 outperformed every model on every measure.
What the full Hy4 release added
Tencent released the full Hy4 model on November 4, following the Hy4 preview announced in August.1
5 Reporting on the launch says the full version scaled up pre-training and post-training and added visual understanding for the first time. It also reported improvements in instruction following, long-document understanding, and long-horizon task execution, alongside the ability to use tools and operate software.
8
30
Those are reported improvements, not quantified comparisons in the supplied excerpts. The available material does not establish how much training data changed or provide specific measured gains in long-context comprehension or agent performance.
Studio Global AI
Continue your research
This page includes a source-backed answer you can continue inside Studio Global.
People also ask
What is the short answer to "Tencent Expands Hy4 Capacity 10× After WorkBuddy Users Hit Limits"?
Tencent said it expanded Hy4 inference capacity tenfold and raised WorkBuddy’s free quota after users hit first day limits.
What are the key points to validate first?
Tencent said it expanded Hy4 inference capacity tenfold and raised WorkBuddy’s free quota after users hit first day limits. Hy4 scored 35 on Artificial Analysis’s Intelligence Index. Agent Arena’s initial announcement placed it 14th overall, while a later report lists 13th.