Alibaba Qwen's 2026-05 multimodal model. 1M context, text + vision + video input, strong coding and tool use at ~1/3 the price of Qwen3.7 Max.
Qwen3.7-Plus is Alibaba Cloud Model Studio's balanced Qwen3.7 tier: a 1M-context, OpenAI-compatible model for text, image, and video input. Alibaba recommends it as the default Plus option for chatbots, office productivity, content generation, summarization, document processing, and coding tools when teams want lower cost than qwen3.7-max but still need thinking mode, function calling, built-in tools, structured output, and long-context retrieval.
TheRouter exposes the model as qwen/qwen3.7-plus with the same OpenAI-style Chat Completions path as other chat models. Runtime configuration lives in standard-models.yaml: 1,000,000-token context, 65,536 max completion tokens, text+image+video input, text output, and support for temperature, max_tokens, top_p, tools, tool_choice, response_format, and stop. This curated module does not override those live fields; it supplies the source ledger and integration guidance around them.
The trade-off is deliberate: Qwen3.7-Plus is not the maximum reasoning tier and it is not open weight. Use it when multimodal input, long context, and cost discipline matter more than the absolute qwen3.7-max reasoning ceiling. For code-agent migrations from qwen3-coder-plus, Alibaba names qwen3.7-plus as the replacement path, but it is a general-purpose multimodal model rather than a dedicated coding checkpoint, so production teams should regression-test repository and tool-call workloads before switching traffic.
| Type | Rate |
|---|---|
| Input | $0.432 / 1M tokens |
| Output | $1.73 / 1M tokens |
| Stable alias | qwen3.7-plus → qwen3.7-plus-2026-05-26alibabacloud.com ↗ | verified |
| Context window | 1,000,000 tokensalibabacloud.com ↗ | verified |
| Modalities | Text, image, and video input; text outputventurebeat.com ↗ | to verify |
| Thinking mode | Supported; Alibaba documents enable_thinking and reasoning.effort controls for Qwen3+ modelsalibabacloud.com ↗ | verified |
| Tooling | Function calling, built-in tools, and structured output are listed as supported for qwen3.7-plusalibabacloud.com ↗ | verified |
| Upstream list pricing — International ≤256K | $0.40 input / $1.60 output per 1M tokens; thinking tokens billed at output ratealibabacloud.com ↗ | verified |
| Upstream list pricing — International 256K–1M | $1.20 input / $4.80 output per 1M tokens; context-cache discount documented upstreamalibabacloud.com ↗ | verified |
| TheRouter listed pricing | $0.32 input / $1.28 output per 1M tokens in standard-models.yaml at curation timetherouter.ai ↗ | verified |
| License / weights | Closed weights; API-only commercial access in reviewed public sourcesventurebeat.com ↗ | to verify |
| Training cutoff | Not publicly disclosed | unknown |
| Benchmark | Distribution | Score | Source |
|---|---|---|---|
Terminal Bench 2.0-Terminus Secondary coverage attributes this score to Alibaba's Qwen3.7-Plus benchmark chart and compares it with DeepSeek-V4-Pro Max at 67.9 and Gemini-3.1 Pro at 63.5. Treat as medium confidence until Alibaba publishes a static benchmark table that can be independently retrieved. | 70.3 | venturebeat.com ↗ | |
ScreenSpot Pro Reported by VentureBeat from Alibaba's Qwen3.7-Plus benchmark chart; useful directional evidence for UI/screenshot understanding, but not an official machine-readable leaderboard row. | 79.0 | venturebeat.com ↗ | |
Public Qwen3.7-Plus benchmark table Alibaba's accessible Model Studio documentation lists capabilities and pricing for qwen3.7-plus, but the public docs reviewed during this pass did not expose a full official table for SWE-bench, MMMU, MMLU, or LiveCodeBench. Do not infer qwen3.7-max numbers for Plus. | Not publicly disclosed as a retrievable official table | alibabacloud.com ↗ |
Use the global api.therouter.ai endpoint shown below for new integrations; the legacy China accelerated endpoint is retired.
curl https://api.therouter.ai/v1/chat/completions -H "Content-Type: application/json" -H "Authorization: Bearer $THE_ROUTER_API_KEY" -d '{
"model": "qwen/qwen3.7-plus",
"messages": [
{"role": "user", "content": "Summarize the key points from this input."}
]
}'Use Qwen3.7-Plus through TheRouter's OpenAI-compatible Chat Completions endpoint for long-context chat, document analysis, and code-review prompts.
curl https://api.therouter.ai/v1/chat/completions \
-H "Authorization: Bearer $THEROUTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "qwen/qwen3.7-plus",
"messages": [
{"role": "system", "content": "You are a senior engineering reviewer."},
{"role": "user", "content": "Review this migration plan and identify the top operational risks."}
],
"temperature": 0.2,
"top_p": 0.8,
"max_tokens": 2048
}'TheRouter's related operational note connects qwen3.7-plus to DashScope routing decisions for coding agents. Read it alongside this model page when you need fallback policy, region behavior, or migration rationale rather than raw model specs.
TheRouter announced both qwen/qwen3.7-max and qwen/qwen3.7-plus under the standard OpenAI-compatible endpoint, positioning Plus as the lower-cost multimodal choice and Max as the stronger reasoning/code-agent tier.
VentureBeat reported Qwen3.7-Plus as a closed/API-only multimodal model with text, image, and video input and cited Alibaba benchmark claims for terminal and screen-understanding tasks. We use those benchmark claims as medium-confidence directional evidence, not as official leaderboard rows.
It is a general-purpose multimodal Qwen3.7 Plus model, not a dedicated Qwen-Coder checkpoint. Alibaba recommends qwen3.7-plus for coding tools because it supports a 1M context window and function calling, and migration content points qwen3-coder-plus users toward qwen3.7-plus. Still, if you relied on code-specialized behavior, run your repository evals before switching production traffic.
TheRouter's model registry lists text+image+video->text for qwen/qwen3.7-plus, and upstream/public coverage describes Qwen3.7-Plus as supporting text, image, and video input. For images, use the OpenAI-compatible message content-array format. For production video workflows, verify current request-size and media-url constraints in the live API before committing UX promises.
Alibaba's international list price is tiered by input length: $0.40/$1.60 per 1M input/output tokens up to 256K input tokens, and $1.20/$4.80 from 256K to 1M. TheRouter's listed runtime pricing at this curation pass is $0.32/$1.28 per 1M tokens. Always treat the live TheRouter model page and your account invoice as the billing source of truth.
Choose Plus when you need multimodal input or lower cost over many large-context calls. Choose Max when the workload is mostly text/code and you need the strongest Qwen reasoning tier. Do not migrate purely on benchmark headlines: run the same prompts, tool calls, and failure cases through both models and compare quality, latency, and total token cost.
No public open-weight release was found for Qwen3.7-Plus during this pass. The reviewed sources describe it as a closed, commercial API model available through Alibaba Cloud Model Studio, Qwen Chat, and downstream API routes such as TheRouter.
| source | URL | retrieved | |
|---|---|---|---|
| Stable alias | alibabacloud.com ↗ | 2026-07-21 | verified |
| Context window | alibabacloud.com ↗ | 2026-07-21 | verified |
| Modalities | venturebeat.com ↗ | 2026-07-21 | to verify |
| Thinking mode | alibabacloud.com ↗ | 2026-07-21 | verified |
| Tooling | alibabacloud.com ↗ | 2026-07-21 | verified |
| Upstream list pricing — International ≤256K | alibabacloud.com ↗ | 2026-07-21 | verified |
| Upstream list pricing — International 256K–1M | alibabacloud.com ↗ | 2026-07-21 | verified |
| TheRouter listed pricing | therouter.ai ↗ | 2026-07-21 | verified |
| License / weights | venturebeat.com ↗ | 2026-07-21 | to verify |
| Training cutoff | — | — | unknown |
| Terminal Bench 2.0-Terminus | venturebeat.com ↗ | 2026-07-21 | to verify |
| ScreenSpot Pro | venturebeat.com ↗ | 2026-07-21 | to verify |
| Public Qwen3.7-Plus benchmark table | alibabacloud.com ↗ | 2026-07-21 | verified |
| Qwen3.7-Plus DashScope routing policy sets the default tier for coding agents | therouter.ai/news ↗ | 2026-07-21 | verified |
| Qwen 3.7 Max and Plus go live on TheRouter | therouter.ai/blog ↗ | 2026-07-21 | verified |
| External coverage frames Qwen3.7-Plus as low-cost multimodal API access | venturebeat.com ↗ | 2026-07-21 | verified |
| Is Qwen3.7-Plus a coding model? | alibabacloud.com ↗ | 2026-07-21 | to verify |
| Does qwen/qwen3.7-plus support image and video input through TheRouter? | therouter.ai ↗ | 2026-07-21 | to verify |
| How is Qwen3.7-Plus priced? | alibabacloud.com ↗ | 2026-07-21 | to verify |
| When should I choose qwen3.7-plus instead of qwen3.7-max? | therouter.ai ↗ | 2026-07-21 | to verify |
| Is Qwen3.7-Plus open source? | venturebeat.com ↗ | 2026-07-21 | to verify |