Now Available: ByteDance Seed 2.0 and Qwen3-Coder-Next — Frontier Reasoning and Code Agents
ByteDance's Seed 2.0 Pro and Alibaba's Qwen3-Coder-Next are now available on the Hanzo AI Gateway — frontier multimodal reasoning and open-source coding agents.
Two models that shipped in the past two weeks are live on the Hanzo AI Gateway.
Seed 2.0 Pro — ByteDance's Frontier Reasoning Model
Released February 14 by ByteDance's Seed Team, Seed 2.0 Pro is a fully multimodal reasoning model (text, image, video input) that competes directly with GPT-5.2 and Gemini 3 Pro on benchmarks. It ranks top-6 on LMSYS Chatbot Arena and top-4 on Vision Arena.
Key specs:
- Full multimodal — text, image, and video input
- State-of-the-art on MathVision and MotionBench
- Strong agentic tool-calling — built for agent workflows
curl https://llm.hanzo.ai/v1/chat/completions \
-H "Authorization: Bearer $HANZO_API_KEY" \
-d '{"model": "bytedance-seed/seed-2.0-mini", "messages": [{"role": "user", "content": "..."}]}'
Also available: Seed 2.0 Mini for high-throughput batch processing.
Qwen3-Coder-Next — Open-Source Coding Agent
Released February 3, Qwen3-Coder-Next is an 80B-parameter MoE model with only 3B active parameters, purpose-built for coding agents. It achieves 70.6% on SWE-bench Verified while running on a single consumer workstation with 46 GB RAM.
Key specs:
- 80B total / 3B active — ultra-sparse MoE
- 256K context window
- Apache 2.0 — fully open for commercial use
- 70.6% SWE-bench Verified — frontier-class code generation
- 10x throughput vs dense models of equivalent quality
curl https://llm.hanzo.ai/v1/chat/completions \
-H "Authorization: Bearer $HANZO_API_KEY" \
-d '{"model": "qwen/qwen3-coder-next", "messages": [{"role": "user", "content": "..."}]}'
The Gateway Advantage
Every model — from ByteDance's proprietary Seed 2.0 to Alibaba's open-source Qwen3-Coder-Next — is accessible through the same API key. No new accounts, no new SDKs. Just change the model name.
100+ models at hanzo.ai.
Read more
Now Available: Gemini 3.1 Pro — Google's 1M-Token Reasoning Model
Google's Gemini 3.1 Pro is now available on the Hanzo AI Gateway — featuring a 1M-token context window and multimodal reasoning across text, images, audio, video, and code.
Now Available: Qwen3.5-35B-A3B and Mercury 2 — Efficient MoE and Frontier Reasoning
Qwen3.5-35B-A3B (Apache 2.0, multimodal MoE) and Inception's Mercury 2 are now available on the Hanzo AI Gateway.
Now Available: Claude Sonnet 4.6 and Qwen3.5-397B — Multimodal MoE at Scale
Claude Sonnet 4.6 and Qwen3.5-397B-A17B are now available on the Hanzo AI Gateway — Anthropic's latest alongside the largest open-source multimodal MoE model.