Repo-aware Qwen3 coder — tools, function calling, 256K context.
qwen3-coder-plus is the coding-specialised flagship of the Qwen3 generation. It is tuned for repo-aware tasks — multi-file refactors, function calling, agentic loops with shell / fs tools, and producing structured patch-style output — with a 256K context window that holds a small repo in a single call.
Bailian publishes tiered pricing for this model based on prompt length; the rate exposed to customers here is the standard short-prompt tier. TheRouter routes between bailian-cn and bailian-sg per request based on cost.
| Type | Rate |
|---|---|
| Input | $1.08 / 1M tokens |
| Output | $5.40 / 1M tokens |
| Stable alias | qwen3-coder-plus → qwen3-coder-plus-2025-09-23www.alibabacloud.com ↗ | verified |
| Available snapshots | qwen3-coder-plus-2025-09-23; qwen3-coder-plus-2025-07-22www.alibabacloud.com ↗ | verified |
| Context window | 1,000,000 tokens upstreamwww.alibabacloud.com ↗ | verified |
| Max input | 997,952 tokens upstreamwww.alibabacloud.com ↗ | verified |
| Max output | 65,536 tokens upstreamwww.alibabacloud.com ↗ | verified |
| Model positioning | Maximum-quality Qwen-Coder choice for highly complex taskswww.alibabacloud.com ↗ | verified |
| Context cache | Supported; upstream bills cache hits at discounted rateswww.alibabacloud.com ↗ | verified |
| Training cutoff | Not publicly disclosed | unknown |
| Benchmark | Distribution | Score | Source |
|---|---|---|---|
SWE-bench Verified Official launch post reports the Qwen3-Coder family / 480B variant achieved SOTA among open-source models without test-time scaling; Alibaba Cloud does not publish a separate numeric benchmark for the managed qwen3-coder-plus-2025-09-23 alias. | State-of-the-art among open-source models for the July 2025 Qwen3-Coder launch; exact Plus snapshot score not separately disclosed | qwenlm.github.io ↗ | |
Agentic Coding / Browser-Use / Tool-Use The Qwen launch announcement describes Qwen3-Coder-480B-A35B-Instruct as setting open-model SOTA on agentic coding, browser-use, and tool-use benchmarks. Treat this as family evidence for Plus unless Alibaba publishes Plus-specific evaluations. | Comparable to Claude Sonnet 4 at launch; exact managed Plus score not publicly disclosed | qwenlm.github.io ↗ | |
Public Plus-specific benchmark disclosure Alibaba Cloud documentation describes positioning and capabilities but does not publish Plus-specific SWE-bench, HumanEval, LiveCodeBench, or Terminal-Bench numbers for the stable alias. | Not publicly disclosed | 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-coder-plus",
"messages": [
{"role": "user", "content": "Summarize the key points from this input."}
]
}'Use Qwen3-Coder-Plus through TheRouter's OpenAI-compatible Chat Completions endpoint for code generation, review, and refactoring prompts.
curl https://api.therouter.ai/v1/chat/completions \
-H "Authorization: Bearer $THEROUTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "qwen/qwen3-coder-plus",
"messages": [{"role": "user", "content": "Refactor this TypeScript module to isolate file-system side effects."}],
"temperature": 0.2,
"top_p": 0.8,
"max_tokens": 4096
}'Alibaba Cloud's model list identifies qwen3-coder-plus as a stable alias currently mapped to qwen3-coder-plus-2025-09-23, while retaining the 2025-07-22 snapshot as a callable version.
Qwen introduced Qwen3-Coder-480B-A35B-Instruct and documented OpenAI-compatible access through Alibaba Cloud Model Studio using the qwen3-coder-plus model name for Qwen Code, Cline, and direct API calls.
It is the managed commercial Plus alias for Qwen3-Coder on Alibaba Cloud Model Studio. Public docs tie the launch to Qwen3-Coder-480B-A35B-Instruct, but Alibaba Cloud now exposes Plus as stable and dated snapshots rather than an open-weight model card. Use the Plus page for managed API behavior and the 480B page for open-weight architecture details.
Alibaba recommends qwen3-coder-next for most scenarios because it balances quality, speed, and cost. Use qwen3-coder-plus when maximum quality is more important than latency or budget, and qwen3-coder-flash when cost/throughput dominates.
Yes. Alibaba documents Qwen-Coder tool calling with OpenAI-compatible Chat Completions. The model returns tool_calls, your application executes the selected function, then sends tool results back for the final answer. Keep tool lists below roughly 20 tools for efficiency.
Not in the official Alibaba Cloud docs reviewed for this curation. Public sources provide family-level Qwen3-Coder claims and model-list specs, but not separate numeric SWE-bench or LiveCodeBench scores for qwen3-coder-plus-2025-09-23.
| source | URL | retrieved | |
|---|---|---|---|
| Stable alias | www.alibabacloud.com ↗ | 2026-06-02 | verified |
| Available snapshots | www.alibabacloud.com ↗ | 2026-06-02 | verified |
| Context window | www.alibabacloud.com ↗ | 2026-06-02 | verified |
| Max input | www.alibabacloud.com ↗ | 2026-06-02 | verified |
| Max output | www.alibabacloud.com ↗ | 2026-06-02 | verified |
| Model positioning | www.alibabacloud.com ↗ | 2026-06-02 | verified |
| Context cache | www.alibabacloud.com ↗ | 2026-06-02 | verified |
| Training cutoff | — | — | unknown |
| SWE-bench Verified | qwenlm.github.io ↗ | 2026-06-02 | to verify |
| Agentic Coding / Browser-Use / Tool-Use | qwenlm.github.io ↗ | 2026-06-02 | to verify |
| Public Plus-specific benchmark disclosure | alibabacloud.com ↗ | 2026-06-02 | verified |
| Qwen3-Coder-Plus stable alias moves to the 2025-09-23 snapshot | alibabacloud.com ↗ | 2026-06-02 | verified |
| Qwen3-Coder launches with managed qwen3-coder-plus access | qwenlm.github.io ↗ | 2026-06-02 | verified |
| Is qwen/qwen3-coder-plus the same as the open Qwen3-Coder-480B model? | www.alibabacloud.com ↗ | 2026-06-02 | to verify |
| Which Qwen-Coder model should I choose: plus, next, or flash? | www.alibabacloud.com ↗ | 2026-06-02 | to verify |
| Does Qwen3-Coder-Plus support function calling? | www.alibabacloud.com ↗ | 2026-06-02 | to verify |
| Are Plus-specific benchmark scores public? | www.alibabacloud.com ↗ | 2026-06-02 | to verify |