# Hy3 vs Nanbeige4.2-3B (slim)

> Hy3 scores 62.6 (B) to Nanbeige4.2-3B's 61 (C) for open-weight models you run yourself. Prices, MCP, x402, uptime and agent notes side by side.

- Full: https://www.anchorterminal.com/compare/hy3-vs-nanbeige-4-2-3b.md (~9,300 tokens) · this version ~780 tokens · JSON https://www.anchorterminal.com/compare/hy3-vs-nanbeige-4-2-3b.json · canonical https://www.anchorterminal.com/compare/hy3-vs-nanbeige-4-2-3b
- Index: https://www.anchorterminal.com/llms.txt · API: https://www.anchorterminal.com/api/v1/index.json · MCP: https://www.anchorterminal.com/mcp · Updated: 2026-11-10
- Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data)

Hy3 scores 62.6 (B) on agent readiness against Nanbeige4.2-3B's 61 (C), and leads in 1 of 4 scored categories. Nanbeige4.2-3B leads on reach, agent interface and documentation and upkeep. Both do open-weight models you run yourself.

- Hy3: B 62.6, rank #28 of 160 open-weight models · https://www.anchorterminal.com/tools/hy3.min.md
- Nanbeige4.2-3B: C 61, rank #78 of 160 open-weight models · https://www.anchorterminal.com/tools/nanbeige-4-2-3b.min.md
- Hy3, good for Teams with an 8-GPU server who want an agent-tuned MoE model under plain Apache 2.0 Ahead on Risk, 60 against 55.
- Nanbeige4.2-3B, good for Local agents that need tool calling and a long context on a laptop or a small GPU Ahead on Reach, 85 against 70; Agent interface, 85 against 75; Documentation and upkeep, 72 against 56.

| Category | Hy3 | Nanbeige4.2-3B |
| --- | --- | --- |
| Capability (35%) | pending | pending |
| Reach (25%) | 70 | 85 |
| Risk (20%) | 60 | 55 |
| Agent interface (10%) | 75 | 85 |
| Documentation and upkeep (10%) | 56 | 72 |

| Fact (where they differ) | Hy3 | Nanbeige4.2-3B |
| --- | --- | --- |
| Vendor | Tencent (Hy team) | Nanbeige LLM Lab (BOSS Zhipin) |
| Licence | Apache-2.0 | Apache-2.0 (Hub metadata only; the repository has no LICENSE file) |
| Last release | 2026-07-20 | 2026-09-11 |
| Agent reviews | 2.8/5 (4) | none |

Why the edge:
- Documentation and upkeep, Nanbeige4.2-3B +16: Hy3 56 (The model card earns 11, covering intended use, architecture, parameters and context but not limitations, languages or a knowledge cut-off.); Nanbeige4.2-3B 72 (Model card 17, Intended use, limitations, the architecture, the parameters (3B outside the embeddings), the context and the languages (in the metadata) are on t…)
- Reach, Nanbeige4.2-3B +15: Hy3 70 (Memory earns 5, because the smallest build of 4-bit precision or better is AngelSlim's GPTQ-Int4 of 168.5 GB, about 202 GB with the fifth added, which is within…); Nanbeige4.2-3B 85 (Memory 30, the smallest build from Nanbeige is the FP8 repository (the 8-bit GPTQ one is the same size), 5.19 GB, and 5.19 GB × 1.2 = 6.2 GB, within 16 GB.)
- Agent interface, Nanbeige4.2-3B +10: Hy3 75 (The chat template earns 20, a chat_template.jinja in the repository.); Nanbeige4.2-3B 85 (Chat template 20, A Jinja chat template in tokenizer_config.json (7,961 characters), with enable_thinking, preserve_thinking, tools and tool_call_format switche…)
- Risk, Hy3 +5: Hy3 60 (File safety earns the full 25, safetensors only with no pickle files, and the documented route runs vLLM or SGLang from their own code with a Jinja template, so…); Nanbeige4.2-3B 55 (File safety 15, safetensors only, no .bin, .pt or .pth files (25).)