Hy3 by Tencent (via OpenRouter)
295B MoE open-weight model (21B active, 192 experts). GPQA Diamond 90.4, SWE-bench Verified 78.0. Apache 2.0.
How to use Hy3
295B open-weight model from Tencent, free for ~2 weeks on OpenRouter. Top GPQA Diamond (90.4) and SWE-bench Verified (78.0). OpenAI-compatible via OpenRouter.
Quickstart
curl https://openrouter.ai/api/v1/chat/completions \
-H "Authorization: Bearer $KEY" \
-d '{"model":"tencent/hy3:free","messages":[{"role":"user","content":"hi"}]}'Frequently asked
Is Hy3 free?
Yes. Tencent (via OpenRouter) offers it as Free API tier with these limits: Free for ~2 weeks from 2026-07-07 via OpenRouter (tencent/hy3:free); standard OpenRouter rate limits apply. No credit card is required.
Can I use Hy3 commercially?
Yes, commercial use is allowed under the Apache-2.0 license. Verify the current license or terms before shipping.
How do I start using Hy3?
Get a free API key from the provider dashboard. Set the base URL to https://openrouter.ai/api/v1 and the model to tencent/hy3:free. 295B open-weight model from Tencent, free for ~2 weeks on OpenRouter. Top GPQA Diamond (90.4) and SWE-bench Verified (78.0). OpenAI-compatible via OpenRouter.
Building with this model?
We wire free-tier and open-weight models into production stacks: evals against your real inputs, failover, cost caps.
Free monthly update
The monthly free-models update
What is newly free, what got rate-limited, and what to switch to. One email a month. Unsubscribe anytime.
Related free models
- Kimi K2.6 (Ollama Cloud) · Free tier
- GLM 4.6 (Z.ai) · Self-host free
- Qwen3.7 Max (Qwen) · Free tier
- Inference API (Hugging Face) · Rate-limited · 1000s of models
- bytedance-seed/dola-seed-2.0-pro:free (Kilo Code) · ~200 req/hr
- Gemini 3.5 Flash (Google Gemini) · 15 RPM, 1,500 RPD