Independent evaluator Artificial Analysis published Tencent Hunyuan Hy4 results on the Intelligence Index v5 on November 5 (08:28 UTC): a composite score of 35, tying Grok 4.7 (xhigh) and one point behind Claude Haiku 5.5 (max). Among China-based labs it ranks second only to Kimi K3.1 Pro (46). On agentic coding it scores 31% on FrontierCode 2.0 and 37% on Terminal-Bench 4.0; cost per Intelligence Index task is about $0.78, roughly one-tenth of same-score Grok 4.7.
Key Takeaways
- ✓Intelligence Index v5 composite 35: ties Grok 4.7 (xhigh), one point behind Haiku 5.5 (max, 36); No.2 China-lab model behind Kimi K3.1 Pro (46)
- ✓FrontierCode 2.0 at 31% (leads China labs; DeepSeek V4.1 Flash 1017 max 28%, Kimi K3.1 Pro 25%); Terminal-Bench 4.0 at 37%
- ✓AA-Omniscience v1.1 accuracy 50% but hallucination rate 89% (Kimi K3.1 Pro hallucination 37%); HLE-Diamond 33%
- ✓About $0.78 per Intelligence Index task: ~10x cheaper than same-score Grok 4.7, ~13x cheaper than Qwen3.8 Max (34), nearly 2x Haiku 5.5
- ✓Pricing about $0.834/$2.501 per 1M input/output tokens, cache reads $0.042; 1M context; text and image input
Key Decision Metrics at a Glance
Turn your technical choice into a development budget
Compare 40 dev plans & simulate token costs vs $20/mo subscriptions
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
Core Background and Industry Pain Points
Tencent Hunyuan released Hy4 GA on November 4 (first vision and computer-use support), mostly with vendor claims and an internal finance eval. Developers still needed an independent landing on the shared Intelligence Index. Artificial Analysis posted Hy4's v5 results on X and its model page at 08:28 UTC on November 5. A prior aicoder news row covers the Hy4 launch; this row covers only the AA independent scores.
Architecture Highlights and Underlying Mechanisms
Intelligence Index v5.0.1 weights knowledge work, coding, science and general at 25% each across nine evals including AA-Briefcase, AutomationBench-AA, FrontierCode 2.0, Terminal-Bench 4.0, AA-Omniscience v1.1 and HLE-Diamond. Per AA, Hy4 takes text and image input, outputs text, and has a 1M context; the AA page still labels it proprietary while weights are not yet public. Pricing is about $0.834/$2.501 per 1M input/output tokens and $0.042 for cache reads.
Benchmarks and Measured Results
AA-reported: composite 35, tying Grok 4.7 (xhigh) and one point behind Haiku 5.5 (max); No.2 among China-based labs behind Kimi K3.1 Pro (46). FrontierCode 2.0 is 31% (vs Claude Fable 5.1 max 32%, DeepSeek V4.1 Flash 1017 max 28%, Kimi 25%); Terminal-Bench 4.0 is 37%. AA-Omniscience accuracy is 50% with an 89% hallucination rate; HLE-Diamond is 33%. Cost per Intelligence Index task is about $0.78. The run produced about 160M output tokens, on the verbose side.
Developer Playbook and Getting Started
Hy4 is live on Tencent Cloud TokenHub and OpenRouter (tencent/hy4); Tencent says weights and system cards are coming soon, with a cheaper Hy4 Flash on the way. For agentic coding, lean on FrontierCode and Terminal-Bench; for knowledge work, treat the 89% hallucination rate as a hard warning and double-check critical facts. API pricing matches the preview tier; WorkBuddy and CodeBuddy have short free windows per Tencent's site.
Benchmark side-by-side against alternatives, or calculate monthly token cost vs subscription break-even.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.