Hyper, OpenCode, Kilo, and Vals add Qwen 3.7 Plus support within 72 hours

• Within about 48 hours of Qwen 3.7 Plus shipping, opencode's Go announcement, kilocode's Kilo post, ValsAI's ranking post, and vercel_dev's AI Gateway post had already added access, rankings, or promo availability around the new model.
• According to Alibaba_Qwen's launch thread, Qwen 3.7 Plus is a multimodal agent model built for GUI and CLI work, and Vercel's model page lists a 1M token context window.
• opencode pitched Qwen3.7 Plus in its Go plan as text plus image with 1M context and lower cost than 3.6, while kilocode marketed the same release as a screen-reading, GUI-navigating coding agent.
• Public eval coverage showed up just as fast: ValsAI placed Qwen 3.7 Plus at #13 on its index, and ValsAI's follow-up specs post said the eval run used temperature 0.7 with a 66k max output token cap.

You can read Qwen's launch post, check Vercel's AI Gateway changelog, browse Hyper's product page, and inspect the live Vals model entry. The interesting bit is how quickly the rollout split by use case: terminal coding in Go, free trials in Kilo and AI Gateway, and almost immediate placement in a public agent leaderboard.

According to Alibaba_Qwen's launch thread, the model combines vision and language in one agent foundation, with unified GUI and CLI operation, coding and productivity workflows, and cross-harness generalization. Vercel's model page adds the core deployment details that kept showing up in tool announcements: 1M context, vision input, and a position as the successor to Qwen3.6-Plus.

That gave downstream tools a clean story to package in different ways. opencode reduced it to text, image, 1M context, and lower price than 3.6, while kilocode's Kilo post emphasized screen reading, GUI navigation, and code generation from visual references.

The fastest pickup happened inside coding-agent surfaces. In opencode's Go announcement, the model landed as part of a hosted plan with the short pitch technical users actually care about: multimodal input, 1M context, and a cheaper step from 3.6.

kilocode's Kilo post used a different angle. It framed Qwen 3.7 Plus as a multimodal agent that can read screens, navigate GUIs, work across GUI and CLI, and generate code from visual references. The attached model picker shows how these launches now land in practice: not as abstract API support, but as one more selectable agent model inside an existing tool.

Not every integration was for Plus. charmcli's Hyper post added Qwen 3.7 Max to Hyper, Charm's inference service for coding agents, after what it described as a week-long check for performance and zero data retention. The Hyper landing page makes the platform positioning explicit: model behavior as shipped by the lab, coding-agent optimization, and privacy controls for teams.

On the Plus side, vercel_dev's AI Gateway post put both Qwen 3.7 Plus and Max behind AI Gateway with temporary free access for paid users through June 4 at noon PT. AskVenice also put Qwen 3.7-Plus live in Venice, which made three separate hosted access points appear within roughly two days of launch.

The other fast-moving surface was public eval infrastructure. ValsAI moved Qwen 3.7 Plus straight to #13 on the Vals Index, two spots above Qwen 3.6 Plus.

ValsAI's follow-up specs post is more useful than the rank tweet by itself because it exposes the run configuration: 1M context, temperature 0.7, support for text, image, and video output, plus a 66k max output token limit. The live Vals model page also timestamps the entry to June 1, the same day as Alibaba's own release.

The launch wave also came with short promo windows, which is a pretty direct way to force first-run usage. kilocode made Qwen 3.7 Plus free in Kilo for a limited time, and vercel_dev did the same for both Plus and Max on AI Gateway through a specific June 4 cutoff.

That matters to the shape of the rollout more than the pricing itself. Instead of waiting for a slow API adoption curve, Alibaba's partners pushed the model into existing coding-agent and gateway products with immediate try-it-now access.