Skip to content

fix(llm_qwen): correct stale and invalid DashScope model profiles #1712

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
chinesepowered wants to merge 4 commits into
base: develop
Choose a base branch
from
Open

fix(llm_qwen): correct stale and invalid DashScope model profiles #1712

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
543003d
fix(llm_qwen): correct stale and invalid DashScope model profiles
chinesepowered Jul 28, 2026
0d6d983
fix(llm_qwen): drop unverified qwen3.8 preview, wire base_url everywhere
chinesepowered Jul 28, 2026
eaea690
docs(llm_qwen): scope the base_url availability claim to live profiles
chinesepowered Jul 28, 2026
9a9e6e3
docs(llm_qwen): correct the discovery fallback description
chinesepowered Jul 29, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
64 changes: 52 additions & 12 deletions nodes/src/nodes/llm_qwen/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@ A RocketRide LLM node that connects Alibaba Cloud Qwen models to a pipeline via

Provides Qwen chat completions to the pipeline. Used primarily as an `llm` invoke connection by agents and other nodes that need an LLM, and can also be used directly via lanes.

Uses **LangChain's `ChatOpenAI`** client pointed at DashScope's OpenAI-compatible endpoint. The regional endpoint is resolved from the `region` field at startup. Temperature is fixed at `0`, and `max_tokens` is taken from the profile's `modelOutputTokens`.
Uses **LangChain's `ChatOpenAI`** client pointed at DashScope's OpenAI-compatible endpoint. The endpoint is resolved at startup from the `base_url` field if set, otherwise from the `region` field. Temperature is fixed at `0`, and `max_tokens` is taken from the profile's `modelOutputTokens`.

When the node configuration is validated, the node performs a live 1-token test request against the API to verify the key, model, and region actually work. Failures surface as configuration warnings with the provider's error message.

Expand All @@ -29,27 +29,52 @@ The main setting is the **profile** (model selection, default `qwen-flash`). Eac
| `profile` | enum, `qwen-flash` | Qwen AI model selection (see profiles below, or `custom`) |
| `apikey` | string | DashScope API key. Must start with `sk-`. |
| `region` | enum, `us` | DashScope regional endpoint: `us`, `intl`, or `cn` (see regions below) |
| `base_url` | string, empty | Optional endpoint override, wins over `region` (see regions below) |
| `model` | string | Qwen model name (custom profile only) |
| `modelTotalTokens` | number | Maximum context length in tokens (custom profile only, must be > 0) |

---

## Profiles

| Profile | Model | Context tokens | Output tokens |
|---------------------------------|---------------------------------|----------------|---------------|
| Qwen Flash *(default)* | `qwen-flash` | 131,072 | 4,096 |
| Qwen Plus | `qwen-plus` | 1,000,000 | 32,768 |
| Qwen2.5 72B Instruct | `qwen-2.5-72b-instruct` | 32,768 | 16,384 |
| Qwen2.5 7B Instruct | `qwen-2.5-7b-instruct` | 32,768 | 32,768 |
| Qwen2.5 Coder 32B Instruct | `qwen-2.5-coder-32b-instruct` | 32,768 | 4,096 |
| Qwen-Max | `qwen-max` | 32,768 | 8,192 |
| Qwen Plus 0728 | `qwen-plus-2025-07-28` | 1,000,000 | 32,768 |
| Qwen Plus 0728 (thinking) | `qwen-plus-2025-07-28:thinking` | 1,000,000 | 32,768 |
| Qwen-Turbo | `qwen-turbo` | 131,072 | 8,192 |
Profiles come in two kinds.

**Stable aliases** always resolve to DashScope's current snapshot for their tier, so they do not go stale as new generations ship. Prefer these unless you need to pin a specific release:

| Profile | Model | Context tokens | Output tokens |
|---------------------------|---------------|----------------|---------------|
| Qwen Flash *(default)* | `qwen-flash` | 131,072 | 4,096 |
| Qwen Plus | `qwen-plus` | 1,000,000 | 32,768 |
| Qwen Max | `qwen-max` | 32,768 | 8,192 |
| Qwen Turbo | `qwen-turbo` | 131,072 | 8,192 |

**Pinned releases** name a specific model version:

| Profile | Model | Context tokens | Output tokens |
|------------------------|------------------------|----------------|---------------|
| Qwen3.8 Max (preview) | `qwen3.8-max-preview` | 983,616 | 131,072 |
| Qwen3.7 Max | `qwen3.7-max` | 1,000,000 | 65,536 |
| Qwen3.7 Plus | `qwen3.7-plus` | 1,000,000 | 65,536 |
| Qwen3.6 Flash | `qwen3.6-flash` | 1,000,000 | 65,536 |
| Qwen Plus 0728 | `qwen-plus-2025-07-28` | 1,000,000 | 32,768 |
Comment thread
coderabbitai[bot] marked this conversation as resolved.
Outdated

Choose `custom` to set the model name and context length manually.

> **Qwen3.8 Max is a preview model.** Access is limited and it may be served from an endpoint other than the standard compatible-mode hosts. If the profile fails to validate, set `base_url` to the endpoint shown in your Alibaba Cloud console.

### Deprecated profiles

These remain selectable so saved pipelines keep loading, but DashScope rejects their model IDs. They were introduced by OpenRouter fallback discovery in the model sync and carry OpenRouter/HuggingFace IDs rather than DashScope ones. Migrate to a profile above.

| Profile | Model | Why it fails |
|----------------------------|---------------------------------|-----------------------------------------------------|
| Qwen2.5 72B Instruct | `qwen-2.5-72b-instruct` | DashScope uses `qwen2.5-72b-instruct` |
| Qwen2.5 7B Instruct | `qwen-2.5-7b-instruct` | DashScope uses `qwen2.5-7b-instruct` |
| Qwen2.5 Coder 32B Instruct | `qwen-2.5-coder-32b-instruct` | DashScope uses `qwen2.5-coder-32b-instruct` |
| Qwen Plus 0728 (thinking) | `qwen-plus-2025-07-28:thinking` | `:thinking` is OpenRouter variant syntax |

DashScope has no `:thinking` model variants — reasoning is controlled with the `enable_thinking` request parameter instead.

---

## Regions
Expand All @@ -64,6 +89,8 @@ Choose `custom` to set the model name and context length manually.

The default is `us`. An unrecognised value falls back to the US endpoint.

Setting `base_url` overrides this table entirely, which is how you reach an endpoint Alibaba Cloud serves outside the three hosts above. Leave it empty to use the regional endpoint.

Note: DashScope API keys are not interchangeable between regions. A key issued for one region will fail authentication against another region's endpoint.

---
Expand All @@ -87,9 +114,22 @@ Rate-limit and connection errors are classified as retryable by the shared chat

---

## Keeping the model list current

Profiles are maintained by the model sync tool, see [tools/sync_models](../../../../tools/sync_models/README.md):

```bash
python tools/sync_models/src/sync_models.py --provider llm_qwen --enable-discovery --apply
```

Discovery requires `ROCKETRIDE_QWEN_KEY`. Without it the sync falls back to OpenRouter, which lists HuggingFace-style IDs that DashScope does not accept — the source of the deprecated profiles above. The stable aliases are listed in `protected_profiles` so a non-authoritative source cannot deprecate them.

---

## Upstream docs

- [DashScope API reference](https://help.aliyun.com/zh/dashscope/)
- [Alibaba Cloud Model Studio models](https://www.alibabacloud.com/help/en/model-studio/models)

---

Expand Down
10 changes: 7 additions & 3 deletions nodes/src/nodes/llm_qwen/qwen_client.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -61,9 +61,13 @@ def __init__(self, provider: str, connConfig: Dict[str, Any], bag: Dict[str, Any
if not apikey or not apikey.startswith('sk-'):
raise ValueError('Invalid DashScope API key. Key must start with "sk-".')

# Resolve regional endpoint
region = config.get('region', 'us')
base_url = DASHSCOPE_REGIONS.get(region, DASHSCOPE_REGIONS['us'])
# Resolve the endpoint, an explicit base_url wins over the regional map
# so a profile can reach a host outside the compatible-mode endpoints
base_url = (config.get('base_url') or '').strip()

if not base_url:
region = config.get('region', 'us')
base_url = DASHSCOPE_REGIONS.get(region, DASHSCOPE_REGIONS['us'])

# Get the llm using OpenAI-compatible endpoint
self._llm = ChatOpenAI(
Expand Down
Loading
Loading