AI tool comparison
Tencent Hy3 Preview vs Ternary Bonsai
Which one should you ship with? Here is the side-by-side panel verdict, pricing read, reviewer split, and community vote comparison.
AI Models
Tencent Hy3 Preview
295B MoE open weights — China's most efficient frontier model yet
75%
Panel ship
—
Community
Paid
Entry
Tencent open-sourced Hy3 Preview on April 23, 2026 — the first model to emerge from the company's rebuilt AI infrastructure, and its most credible challenge to frontier closed models to date. With 295 billion total parameters but only 21 billion active at inference time (plus 3.8B MTP layer parameters), it's a Mixture-of-Experts architecture that punches far above its compute weight. The model supports up to 256K context and is available via Hugging Face, ModelScope, and GitCode under the Tencent Hy Community License. On coding benchmarks, Hy3 scores 74.4% on SWE-bench Verified, 54.4% on Terminal-Bench 2.0, and 67.1% on BrowseComp — placing it firmly in the same tier as top models from Anthropic and OpenAI. Tencent claims a 40% efficiency improvement over its predecessor Hunyuan models, and pricing through Tencent Cloud TokenHub is aggressive: RMB 1.2 per million input tokens. A free two-week window at launch via OpenRouter made it widely accessible immediately. The model was led by a team that includes former OpenAI researchers and has already been deployed across Tencent's core products — WeChat, Yuanbao, and QQ. That production integration is a meaningful signal: this isn't a benchmark vanity release. For developers who need a powerful, cost-efficient reasoning and agentic model with actual open weights, Hy3 Preview is one of the most interesting drops of April 2026.
Open Source Models
Ternary Bonsai
1.58-bit LLMs that run at 82 tok/s on M4 Pro and on your iPhone
75%
Panel ship
—
Community
Free
Entry
PrismML's Ternary Bonsai is a family of aggressively quantized language models that take the BitNet concept to its logical extreme. Each weight is constrained to one of three values — {-1, 0, +1} — with a shared FP16 scale factor per 128-weight group. No higher-precision escape hatches, no hybrid layers. The result is a 9x reduction in memory footprint versus standard 16-bit models. The numbers are striking: the 8B model fits in 1.75 GB and hits 82 tokens per second on an M4 Pro. More impressively, it runs at 27 tokens per second on an iPhone 17 Pro Max — fast enough for real-time conversation on-device. The 8B variant scores 75.5 average across standard benchmarks, outperforming many models that are 9-10x larger. The 4B and 1.7B variants push further into mobile-optimized territory. All three models are released under the Apache 2.0 license, available on Hugging Face and GitHub, and integrated into the Locally AI iOS app for immediate on-device deployment. For developers building privacy-sensitive applications or anyone tired of paying cloud inference costs, Ternary Bonsai offers a compelling on-device alternative that doesn't require a beefy GPU.
Reviewer scorecard
“21B active params with 295B total — this is genuinely practical to deploy on reasonable hardware while matching models 10x the inference cost. The 256K context and strong SWE-bench score make it a legitimate option for agentic coding pipelines. I'd use this today.”
“82 tokens per second on M4 Pro in 1.75 GB is a genuinely impressive engineering achievement. For local tooling, code assistants, or any latency-sensitive workload where I don't want cloud round-trips, this hits a sweet spot that larger quantized models miss. Apache 2.0 means I can embed it in commercial apps without legal headaches.”
“The Tencent Hy Community License is not Apache 2.0 or MIT — read it carefully before using this in production. There are usage restrictions that could bite commercial deployments. Also, benchmark scores look great, but independent evals of Chinese labs' models have historically diverged from self-reported numbers.”
“A 75.5 benchmark average sounds good until you compare it against 8B models quantized with GGUF Q8 — which score similarly and have years of tooling, community support, and production deployments behind them. The 9x memory savings matter on constrained devices but less so on any machine with 16GB+ RAM. Niche but real use case.”
“The MoE efficiency race is the actual story here — we're getting frontier-class capability at a fraction of the activation cost. Hy3 is proof that the compute-vs-capability Pareto frontier keeps moving. Open weights with real deployment signals (WeChat at scale) is a combination that matters.”
“On-device AI at 27 tokens per second on a phone is the inflection point that makes LLMs a platform primitive rather than a cloud service. Once inference is this cheap and fast on commodity hardware, the entire economic model of AI-as-API-call collapses. Ternary quantization is an early signal of where efficiency research is heading.”
“Strong visual coding capabilities and multimodal understanding make this genuinely useful for design-to-code workflows. The health image analysis and product comparison use cases already deployed in Yuanbao show real-world creative utility beyond pure benchmark games.”
“The prospect of running a capable LLM entirely on my iPhone without sending any data to a server is genuinely exciting for creative work with sensitive material. Drafting, editing, and ideation without a cloud subscription or privacy concerns — I'd pay for that, and here it's free.”
Weekly AI Tool Verdicts
Get the next comparison in your inbox
New AI tools ship daily. We compare them before you waste an afternoon.