For over a year, the open-weight model market has coexisted with an awkward limitation: many of the strongest Chinese releases were off-limits to a large portion of interested enterprises. License terms that excluded the European Union, the United Kingdom, and South Korea forced legal teams to kill deployments before engineers could finish evaluations. This affected not only companies headquartered in those regions but any enterprise serving traffic into those areas. For IT teams evaluating open models, the trade-offs became explicit.
Tencent has just removed that barrier. The Hunyuan team released the full version of Hy3, a 295-billion-parameter Mixture-of-Experts (MoE) model with 21 billion active parameters per forward pass, shipped under the permissive Apache 2.0 license, a reversal from the April preview. The open-model community reacted immediately, with researchers on X highlighting the license change as the real headline, and one widely shared post arguing that if the scores hold, Tencent has become an open-source leader. The company stated it will be free on OpenRouter for two weeks.
Ten weeks from preview to product, shaped by 50 internal teams
The April preview of Hy3 was the first model from Tencent's rebuilt pre-training and reinforcement learning infrastructure, released less than three months after the February rebuild. Chief AI Scientist Shunyu Yao framed the early release as a deliberate move to gather feedback from developers and users before the official version. According to the model card, the team collected input from over 50 product teams after the late-April preview, fixed issues in task execution and interaction, and scaled up its post-training pipeline.
Sponsored Protocol
The architecture remains unchanged: 295B total parameters, 21B active per forward pass via top-8 routing across 192 experts, a 3.8B-parameter multi-token prediction (MTP) layer for speculative decoding, and a 256K context window. What changed is behavior. Tencent claims the full release significantly outperforms similar-size models and rivals flagship open-source models with two to five times its parameters.
Blind test favors Hy3 over GLM 5 1 but GLM 5 2 still dominates coding
Tencent's headline evaluation is a blind human study rather than a leaderboard. Arguing that public benchmarks don't tell the full story, the company ran a test with 270 experts across disciplines working on real-world workflows, collecting 312 valid comparisons. Tencent reports Hy3 scored 2.67 out of 4 against GLM-5.1's 2.51, with clear advantages in frontend development, CI/CD, and data and storage work.
The choice of opponent matters. Zhipu AI released GLM-5.2 in mid-June, and Tencent's own benchmark appendix shows GLM-5.2 ahead of Hy3 across the entire agentic coding suite: SWE-bench Verified (84.2 vs. 78.0), SWE-bench Multilingual (83.0 vs. 75.8), Terminal-Bench 2.1 (81 vs. 71.7) and DeepSWE by a wide margin (46.2 vs. 28.0). The blind test targeted the older model; the new one keeps the coding crown.
Sponsored Protocol
GLM-5.2's coding lead is less surprising given size: GLM-5.2 is roughly a 744-billion-parameter MoE with about 40 billion active parameters per token, against Hy3's 295 billion total and 21 billion active. Tencent fields a model with less than half the parameters and nearly half the per-token compute of the one it trails.
Hy3's genuine wins lie elsewhere. On agentic search, it scores 84.2 on BrowseComp and 91.0 on DeepSearchQA, ahead of every open model in Tencent's table and competitive with Claude Opus 4.8 and GPT-5.5. It leads the open field on tool orchestration (79.1 on the public MCP-Atlas set), on agent-harness evaluations like ClawEval, and on long-context retrieval (73.4 on AA-LCR). Taken together, the appendix suggests a model that is arguably the best open-weight choice for search-and-tool-heavy agent workloads, while conceding repository-scale coding to GLM-5.2.
One caveat applies to both wins and losses: nearly all competitor numbers in Tencent's appendix come from Tencent's own test runs. Independent verification from indices like Artificial Analysis is still pending.
Sponsored Protocol
Reliability pitch: hallucination rates cut in half
What makes the release most interesting for enterprise buyers is the set of numbers Tencent chose to emphasize instead of benchmarks. The model card reads less like a leaderboard announcement and more like a production reliability report. In internal evaluations on real-world scenarios, Tencent says Hy3's hallucination rate dropped from 12.5% to 5.4% compared to the preview, and commonsense error rates fell from 25.4% to 12.7% — improvements attributed to fine-grained data cleaning and training constraints built around explicit behavior patterns: answer when grounded, state when evidence is missing, don't conflate sources, don't fabricate data. Multi-turn behavior received similar treatment: the issue rate on internal multi-turn tests fell from 17.4% to 7.9%, and Hy3's score on the open MRCR long-dialogue benchmark jumped from 42.9% to 75.1%.
Tencent also emphasizes consistency across agent scaffolds, reporting SWE-bench variance within a few points whether the model runs inside Claude Code-style harnesses, Cline or KiloCode. That is an underrated property: enterprises rarely control which agent framework their teams standardize on, and a model that only performs in one harness is a hidden integration cost. These are self-reported internal measurements and deserve the same skepticism as any vendor benchmark. But the choice to foreground them signals who Tencent believes its customer is: teams that have been burned by models that demo well and fabricate confidently in production.
Sponsored Protocol
Deployment math: a 295B model in a 744B world on export-compliant silicon
The reliability story connects directly to economics, and here Hy3's coding gap against GLM-5.2 starts to look like a deliberate trade rather than a loss. GLM-5.2 has roughly 744 billion total parameters and about 40 billion active per token; in FP8, its weights alone consume about 744 GB, making an 8x H200 node the practical minimum for production serving. Hy3, at 295B total parameters, carries an FP8 footprint under 300 GB — less than half the memory, with roughly half the active parameters per token driving lower per-request compute. For an organization deciding what to self-host, that is the difference between one heavily-specced node and something far more attainable, with room left for KV cache and batching.
There is a geopolitical wrinkle worth noticing: Tencent's recommended serving configuration targets Nvidia's H20-3e, the memory-boosted variant of the H20, the GPU Nvidia designed specifically to comply with U.S. export restrictions on China. Unlike GLM-5.2, there is no mention of Huawei or Ascend chips. In other words, the model is sized so that eight of the chips Chinese companies can legally buy comfortably serve it at full precision. That constraint-driven design has a convenient side effect for everyone else: a model that runs well on deliberately capped silicon runs even more comfortably on the H100s, H200s and B200s available in Western data centers, via standard vLLM and SGLang deployments with MTP speculative decoding.
Sponsored Protocol
With the Apache 2.0 license — no regional exclusions, no field-of-use restrictions — the enterprise equation becomes clear. GLM-5.2 remains the open-weight choice when coding performance is the only criterion and an 8x H200 budget is available. Hy3 makes its case everywhere else: search and tool-heavy agent workloads, reliability-sensitive applications, and organizations that want frontier-adjacent capability without frontier-scale infrastructure. The open question is whether Western enterprises, now that the license barrier is gone, will treat a Tencent model as a serious candidate at all — or whether the next Artificial Analysis update settles the benchmark debate before procurement gets the chance.
