Fireworks AI 已在其推理基础设施上宣布推出 Qwen 3.7 Plus 和 Qwen 3.7 Max 模型。这些模型专为长周期代理循环而设计，并提供保留推理历史记录以及原生图像/文本输入等功能。Fireworks 强调，用户请求通过其硬件端到端执行，使用授权权重，通过零数据保留和 99.9% 的 SLA 来确保对延迟、吞吐量和数据路径的控制。 AI
影响 增强了代理循环能力，并为 Qwen 模型提供了直接硬件执行。
排序理由 推理提供商发布新 AI 模型的产品。
AI 生成摘要 · Google Gemini · 来自 5 个来源。 我们如何撰写摘要 →
Drop it into what you already use. OpenAI + Anthropic compatible endpoints. Works with Claude Code, Cursor, LangChain, etc. Run it serverless today, and get in touch if you’re interested in early-access for Qwen 3.7 Max or bespoke workloads. Get started: https://t.co/PaZEwfjY5…
Fireworks serves Qwen 3.7 Plus as a true inference provider. Your requests execute end-to-end on our hardware from the licensed weights. No forwarding. Qwen 3.7 Plus (thinking) matches Max on AIME 2025 and delivers 3.55× higher end-to-end throughput than 3.6 Plus.
Built for long-horizon agent loops. Observe → reason → code → act (GUI/CLI) → verify, repeat. Qwen’s own demo ran 11 hours with 10k+ lines of code and 1k+ calls. On Fireworks you get: → reasoning_history="preserved" to persist reasoning context across turns →
Qwen 3.7 Plus is now live on Fireworks. You get the official weights running on our stack. That means full control of latency, throughput, and data path end-to-end, with zero data retention and our 99.9% SLA. Let’s dig in ↓ https://t.co/4JAmGyj9PE
Use state-of-the-art, open-source LLMs and image models at blazing fast speed, or fine-tune and deploy your own at no additional cost with Fireworks AI!