[0genlab]

Summary

Automated sync of the AiHubMix model catalog, generated by aihubmix-canon at cf77bb850390ff8ef71e1dbd5a7cd3ce05445b8f.

The catalog is generated instead of fetched at runtime because a Dify model-provider plugin cannot: the plugin daemon freezes the model list from the yaml shipped inside the package, so a model that is added, renamed, repriced or retired on the gateway only reaches users through a PR like this one.

Scope: models/llm/ — model definition yaml, plus the manifest.yaml version bump, and the test / routing edits below. The registered-id total pinned in tests/test_top100_models.py moves 259 → 257 in this PR as well: that number is pinned by the plugin's own test suite (test_every_new_model_is_registered), so it has to move in the same PR that changes _position.yaml. No plugin code (main.py, provider/, pyproject.toml, uv.lock) is touched.

Full machine-generated report (every add / patch / retire with its reason) is collapsed below. renamed lines in the diff are a git heuristic over the similar-looking yaml skeletons — read the report, not the diff's rename pairing.

Note for reviewers: this description is machine-generated and refreshed on every push to this branch. Editing it by hand is fine — the automation notices (the description carries a marker) and then stops refreshing it instead of overwriting your edit; substantive feedback is still better in the conversation.

Full sync report (machine-generated)

Model sync report

Generated from aihubmix-canon cf77bb850390ff8ef71e1dbd5a7cd3ce05445b8f.

model type	directory	add	regenerate	patch	retire
llm	models/llm/	0	0	0	2
rerank	models/rerank/	0	0	0	0
text-embedding	models/text_embedding/	0	0	0	0
tts	models/tts/	0	0	0	0
speech2text	models/speech2text/	0	0	0	0

Retired (de-listed, gone from the gateway catalog, or upstream model name expired)（2）

• grok-4-fast-non-reasoning — 目录里查无此模型
• grok-4-fast-reasoning — 目录里查无此模型

models/rerank/ — skipped, no context window（2）

cohere-rerank-v5-fast, cohere-rerank-v5-pro

⚠ models/text_embedding/ — needs a human（4）

网关在架但插件里没有。text-embedding 的必填字段（max_chunks（单次批量上限））网关与 canon 都没有，编一个值出来就是在 UI 上
显示一个假参数，所以这里不生成，等人补。

embed-v-4-0, gemini-embedding-2, text-embedding-004, text-search-ada-doc-001

⚠ models/tts/ — needs a human（6）

网关在架但插件里没有。tts 的必填字段（voices[] / default_voice / word_limit / audio_type）网关与 canon 都没有，编一个值出来就是在 UI 上
显示一个假参数，所以这里不生成，等人补。

gemini-3.8-flash-lite-tts, gemini-3.8-flash-tts, gpt-audio-1.5, gpt-realtime-2.1, qwen-audio-3.0-tts-flash, qwen-audio-3.0-tts-plus

⚠ models/speech2text/ — needs a human（4）

网关在架但插件里没有。speech2text 的必填字段（file_upload_limit / supported_file_extensions）网关与 canon 都没有，编一个值出来就是在 UI 上
显示一个假参数，所以这里不生成，等人补。

gpt-4o-transcribe-diarize, gpt-audio-1.5, gpt-live-transcribe, gpt-realtime-2.1

Skipped — no research yet（343）

Live on the gateway but not yet covered by our capability research, so no yaml is generated.

list

AiHubmix-Phi-4-mini-reasoning, AiHubmix-mistral-medium, ByteDance-Seed/Seed-OSS-36B-Instruct, DeepSeek-v3, Doubao-1.5-lite-32k, Doubao-1.5-pro-256k, Doubao-1.5-pro-32k, Doubao-1.5-thinking-pro, Doubao-1.5-vision-pro-32k, Doubao-lite-128k, Doubao-lite-32k, Doubao-lite-4k, Doubao-pro-128k, Doubao-pro-256k, Doubao-pro-32k, Doubao-pro-4k, GPT-OSS-20B, Gryphe/MythoMax-L2-13b, LongCat-Flash-Chat, MiniMax-Text-01, MiniMaxAI/MiniMax-M1-80k, Mistral-large-2407, Pro/THUDM/GLM-4.1V-9B-Thinking, Qwen/QVQ-72B-Preview, Qwen/QwQ-32B, Qwen/QwQ-32B-Preview, Qwen/Qwen2-1.5B-Instruct, Qwen/Qwen2-57B-A14B-Instruct, Qwen/Qwen2-72B-Instruct, Qwen/Qwen2-7B-Instruct, Qwen/Qwen2.5-32B-Instruct, Qwen/Qwen2.5-72B-Instruct, Qwen/Qwen2.5-72B-Instruct-128K, Qwen/Qwen2.5-7B-Instruct, Qwen/Qwen2.5-Coder-32B-Instruct, Qwen2-VL-72B-Instruct, Qwen2-VL-7B-Instruct, Qwen3-235B-A22B-Thinking-2507, THUDM/GLM-4-32B-0414, THUDM/GLM-4-9B-0414, THUDM/GLM-4.1V-9B-Thinking, THUDM/GLM-Z1-32B-0414, THUDM/GLM-Z1-9B-0414, WizardLM/WizardCoder-Python-34B-V1.0, ahm-Phi-3-5-MoE-instruct, ahm-Phi-3-5-mini-instruct, ahm-Phi-3-5-vision-instruct, ahm-Phi-3-medium-128k, ahm-Phi-3-medium-4k, ahm-Phi-3-small-128k, aihub-Phi-4, aihub-Phi-4-mini-instruct, aihub-Phi-4-multimodal-instruct, aihubmix-Codestral-2501, aihubmix-Cohere-command-r, aihubmix-Jamba-1-5-Large, aihubmix-Llama-3-1-70B-Instruct, aihubmix-Llama-3-1-8B-Instruct, aihubmix-Llama-3-2-11B-Vision, aihubmix-Llama-3-2-90B-Vision, aihubmix-Llama-3-70B-Instruct, aihubmix-Mistral-Large-2411, aihubmix-Mistral-large, aihubmix-Mistral-large-2407, aihubmix-command-r-08-2024, aihubmix-command-r-plus, aihubmix-command-r-plus-08-2024, aihubmix-router, alicloud-deepseek-v3.2, alicloud-glm-4.7, alicloud-glm-5, alicloud-kimi-k2-instruct, alicloud-kimi-k2-thinking, alicloud-minimax-m2.5, anthropic-opus-4-6, azure-deepseek-v3.2, azure-deepseek-v3.2-speciale, bai-qwen3-vl-235b-a22b-instruct, baidu/ERNIE-4.5-300B-A47B, cbs-glm-4.7, cerebras-llama-3.3-70b, chatglm_lite, chatglm_pro, chatglm_std, chatglm_turbo, chutesai/Mistral-Small-3.1-24B-Instruct-2503, claude-2, claude-2.0, claude-2.1, claude-3-haiku-20240229, claude-3-haiku-20240307, claude-3-haiku@20240307, claude-3-opus@20240229, claude-3-sonnet-20240229, claude-instant-1, claude-instant-1.2, code-davinci-edit-001, codestral-latest, codex-mini-latest, coding-glm-4.5-air, coding-kimi-k3.1-pro, coding-kimi-k3.1-pro-highspeed, coding-xiaomi-mimo-v2.6-pro, command, command-light, command-light-nightly, command-nightly, command-r, command-r-08-2024, command-r-plus, command-r-plus-08-2024, computer-use-preview, davinci, davinci-002, deepinfra-gemma-4-26b-a4b-it, deepinfra-llama-3.1-8b-instant, deepinfra-llama-3.3-70b-instant-turbo, deepinfra-llama-4-maverick-17b-128e-instruct, deepinfra-llama-4-scout-17b-16e-instruct, deepinfra-nvidia-nemotron-3-nano-30b-a3b2, deepseek-ai/DeepSeek-Coder-V2-Instruct, deepseek-ai/DeepSeek-Prover-V2-671B, deepseek-ai/DeepSeek-R1-Distill-Llama-70B, deepseek-ai/DeepSeek-R1-Distill-Llama-8B, deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B, deepseek-ai/DeepSeek-R1-Distill-Qwen-14B, deepseek-ai/DeepSeek-R1-Distill-Qwen-32B, deepseek-ai/DeepSeek-R1-Distill-Qwen-7B, deepseek-ai/DeepSeek-R1-Zero, deepseek-ai/DeepSeek-V2-Chat, deepseek-ai/DeepSeek-V2.5, deepseek-ai/Janus-Pro-7B, deepseek-ai/deepseek-llm-67b-chat, deepseek-ai/deepseek-vl2, deepseek-r1-distill-llama-70b, deepseek-r1-distill-qianfan-llama-8b, deepseek-v3, doubao-1-5-pro-256k-250115, doubao-1-5-pro-32k-250115, doubao-1-5-thinking-vision-pro-250428, doubao-seed-1-6-250615, doubao-seed-1-6-flash-250615, doubao-seed-1-6-thinking, doubao-seed-1-6-thinking-250615, doubao-seed-1-6-vision-250815, doubao-seed-2-1-pro-260628, doubao-seed-code-preview-latest, ernie-x1-turbo, ernie-x1.1-preview, gemini-2.5-pro-exp-03-25, gemini-2.5-pro-preview-06-05, gemini-2.5-pro-preview-06-05-search, gemini-exp-1114, gemini-exp-1121, gemini-exp-1206, gemini-nano-banana-2.1, gemini-pro, gemini-pro-vision, gemma-3-12b-it, gemma-3-1b-it, gemma-3-27b-it, gemma-3-4b-it, gemma-3n-e4b-it, gemma-7b-it, glm-3-turbo, glm-4, glm-4-flash, glm-4-plus, glm-4.5, glm-4.5-air, glm-4.5-airx, glm-4.5-x, glm-4v, glm-4v-plus, glm-5.3-flashx, glm-5.3-prime, glm-zero-preview, google-gemma-3-12b-it, google-gemma-3-27b-it, google-gemma-3-4b-it, google/gemini-exp-1114, google/gemma-2-27b-it, gpt-3.5-turbo-16k, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-08-06-global, gpt-4o-2024-11-20, gpt-4o-mini, gpt-4o-mini-2024-07-18, gpt-4o-mini-global, gpt-4o-mini-search-preview, gpt-4o-search-preview, gpt-4o-zh, gpt-5.6-sol-pro, gpt-oss-120b, gpt-oss-20b, grok-2-1212, grok-2-vision-1212, grok-3, grok-3-beta, grok-3-fast, grok-3-fast-beta, grok-3-mini, grok-3-mini-beta, grok-3-mini-fast-beta, grok-vision-beta, groq-llama-3.1-8b-instant, groq-llama-3.3-70b-versatile, groq-llama-4-maverick-17b-128e-instruct, groq-llama-4-scout-17b-16e-instruct, kat-dev, kimi-k3.1-pro, kimi-k3.1-pro-highspeed, kimi-thinking-preview, laguna-s-2.1, laguna-xs-2.1, learnlm-1.5-pro-experimental, ling-3.0-flash, llama-3.1-405b-instruct, llama-3.1-405b-reasoning, llama-3.1-70b, llama-3.1-70b-versatile, llama-3.1-8b-instant, llama-3.2-11b-vision-preview, llama-3.2-1b-preview, llama-3.2-3b-preview, llama-3.2-90b-vision-preview, llama-3.3-70b, llama-3.3-70b-instruct, llama2-7b-2048, llama3-70b-8192, llama3-70b-8192(33), llama3-8b-8192, llama3-8b-8192(33), llama3-groq-70b-8192-tool-use-preview, llama3-groq-8b-8192-tool-use-preview, llama3.1-8b, mercury-2.5, meta-llama-3-70b, meta-llama-3-8b, meta-llama/Llama-3.2-90B-Vision-Instruct, meta-llama/llama-3.1-405b-instruct:free, meta-llama/llama-3.1-70b-instruct:free, meta-llama/llama-3.1-8b-instruct:free, meta-llama/llama-3.2-11b-vision-instruct:free, meta-llama/llama-3.2-3b-instruct:free, meta/llama-3.1-405b-instruct, meta/llama3-8B-chat, minimax-m3.1-flash, mistralai/mistral-7b-instruct:free, moonshot-v1-128k, moonshot-v1-128k-vision-preview, moonshot-v1-32k, moonshot-v1-32k-vision-preview, moonshot-v1-8k, moonshot-v1-8k-vision-preview, moonshotai/Kimi-Dev-72B, moonshotai/Moonlight-16B-A3B-Instruct, nvidia-llama-3.1-nemotron-70b-instruct, nvidia-llama-3.3-nemotron-super-49b-v1.5, nvidia-nemotron-3-nano-30b-a3b, nvidia-nemotron-nano-12b-v2-vl, nvidia-nemotron-nano-9b-v2, nvidia/Llama-3_1-Nemotron-Ultra-253B-v1, o1, o1-2024-12-17, o1-global, o1-mini, o1-mini-2024-09-12, o1-preview, o1-preview-2024-09-12, o1-pro, o3-deep-research, o3-global, o3-mini-global, o3-pro-global, qianfan-chinese-llama-2-13b, qianfan-llama-vl-8b, qianfan-qi-vl, qwen-3-235b-a22b-instruct-2507, qwen-3-235b-a22b-thinking-2507, qwen-3-32b, qwen-3.8-27b, qwen-flash, qwen-flash-2025-07-28, qwen-long, qwen-max, qwen-max-0125, qwen-max-longcontext, qwen-mt-plus, qwen-mt-turbo, qwen-plus, qwen-plus-2025-04-28, qwen-plus-2025-07-28, qwen-qwq-32b, qwen-turbo, qwen-turbo-2024-11-01, qwen-turbo-2025-04-28, qwen-turbo-latest, qwen2.5-14b-instruct, qwen2.5-32b-instruct, qwen2.5-3b-instruct, qwen2.5-72b-instruct, qwen2.5-7b-instruct, qwen2.5-coder-1.5b-instruct, qwen2.5-coder-7b-instruct, qwen2.5-math-1.5b-instruct, qwen2.5-math-72b-instruct, qwen2.5-math-7b-instruct, qwen2.5-vl-72b-instruct, qwen3-0.6b, qwen3-1.7b, qwen3-14b, qwen3-235b-a22b-2507, qwen3-235b-a22b-instruct-2507, qwen3-235b-a22b-thinking-2507, qwen3-30b-a3b, qwen3-30b-a3b-instruct-2507, qwen3-30b-a3b-thinking-2507, qwen3-32b, qwen3-4b, qwen3-8b, qwen3-coder-30b-a3b-instruct, qwen3-coder-480b-a35b-instruct, qwen3-coder-flash, qwen3-coder-plus, qwen3-coder-plus-2025-07-22, qwen3-max-preview, qwen3-next-80b-a3b-thinking, sf-kimi-k2-thinking, step-2-16k, stepfun-ai/step3, tencent/Hunyuan-A13B-Instruct, tencent/Hunyuan-MT-7B, text-ada-001, text-babbage-001, text-curie-001, text-davinci-002, text-davinci-003, text-davinci-edit-001, tngtech/DeepSeek-R1T-Chimera, unsloth/gemma-3-12b-it, unsloth/gemma-3-27b-it

Skipped — by alias policy（61）

按既定口径排除，不是「网关有而我们漏同步」。构成：-free 免费档变体 39 个、cc-* Claude Code 路由别名（endpoints 为空，在 Dify 的对话场景里没有意义） 21 个、按 id 点名排除（路由模型，给不出确定的能力面与价格） 1 个。

• -free 免费档变体：agents-a1-free, coding-glm-4.6-free, coding-glm-4.7-free, coding-glm-5-free, coding-glm-5-turbo-free, coding-glm-5.1-free, coding-glm-5.2-free, coding-glm-5.3-flash-free, coding-glm-5.3-free, coding-kimi-k3-free, coding-minimax-m2-free, coding-minimax-m2.1-free, coding-minimax-m2.5-free, coding-minimax-m2.7-free, coding-minimax-m3-free, dots-3-note-preview-free, flint-v1.0-free, gemma-4-26b-a4b-it-free, gemma-4-31b-it-free, glm-4.7-flash-free, intern-s2-free, k2.6-code-preview-free, kimi-for-coding-free, laguna-s-2.1-free, laguna-xs-2.1-free, lfm-2.5-2.6b-free, mimo-v2-flash-free, nemotron-3-nano-omni-30b-a3b-reasoning-free, nemotron-3-ultra-550b-a55b-free, nemotron-3.5-content-safety-free, nemotron-3.5-lightning-free, nemotron-3.5-super-vl-free, north-mini-code-free, qwen3.6-plus-preview-free, qwen3.8-flash-free, qwen3.8-max-free, xiaomi-mimo-v2.6-flash-free, xiaomi-mimo-v2.6-pro-free, xing4.0-29b-a4b-free

• cc-* Claude Code 路由别名（endpoints 为空，在 Dify 的对话场景里没有意义）：cc-MiniMax-M2, cc-deepseek-v3, cc-deepseek-v3.1, cc-doubao-seed-code-preview-latest, cc-ernie-4.5-300b-a47b, cc-glm-4.6, cc-glm-4.7, cc-glm-5, cc-glm-5-turbo, cc-glm-5.1, cc-k2.6-code-preview, cc-kimi-dev-72b, cc-kimi-for-coding, cc-kimi-k2-instruct, cc-kimi-k2-instruct-0905, cc-kimi-k2-thinking, cc-minimax-m2, cc-minimax-m2.1, cc-minimax-m2.5, cc-minimax-m2.5-highspeed, cc-minimax-m3

• 按 id 点名排除（路由模型，给不出确定的能力面与价格）：auto

Release Notes

• Removed 2 model(s) no longer served by the gateway catalog: grok-4-fast-non-reasoning, grok-4-fast-reasoning

Change Type

• Documentation / non-plugin change
• Non-LLM plugin (tools, extensions, datasource, etc.)
• LLM plugin

Screenshots / Videos

Before	After
Model definition yaml only — nothing to screenshot.	New / updated models show up in the Dify model picker.

LLM Plugin Checklist

Areas affected by this change (check all that apply)

• Message flow (system messages, user ↔ assistant turn-taking)
• Tool interaction flow (multi-round usage, Agent App and Agent Node)
• Multimodal input (images, PDFs, audio, video, etc.)
• Multimodal output (images, audio, video, etc.)
• Structured output (JSON, XML, etc.)
• Token consumption metrics
• Other LLM functionality (reasoning, grounding, prompt caching, etc.)
• New models / model parameter fixes

Version

• Bumped top-level version in manifest.yaml (not the one under meta) — this PR moves it 0.0.58 → 0.0.59.
• dify_plugin>=0.3.0,<0.11.0 is declared in pyproject.toml and locked in uv.lock (or kept in requirements.txt for legacy plugins without uv.lock) — SDK docs — this checkout declares >=0.10.2 and locks 0.10.2, inside the supported range; untouched by this PR.

Testing

• Local deployment — Dify version: 1.7.0
• SaaS (cloud.dify.ai)

Why both testing boxes are left unchecked

This automation rewrites model definition yaml; it never boots a Dify instance, so it cannot honestly tick either box. What is verified before this PR is opened:

• the generator's unit tests pin the yaml invariants (feature → capability mapping, per-backend parameter names, pricing unit/currency, retire rules) in aihubmix-canon's make check;
• the plugin's own test suite runs upstream on approval (.github/workflows/pre-pr-check-per-plugin.yaml);
• the diff is yaml + manifest.yaml, plus the pinned registered-id total in tests/test_top100_models.py (259 → 257) — that number is pinned by test_every_new_model_is_registered, so it moves in the same PR that changes _position.yaml. No main.py / provider/ code path is touched.

A human should still confirm the new models behave in the Dify model picker before merging.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota.

An organization owner can choose the repository organization as the billing source for reviews requested by members with a paid Copilot license. Learn about Copilot code review billing. Review the Copilot code review billing settings.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota.

An organization owner can choose the repository organization as the billing source for reviews requested by members with a paid Copilot license. Learn about Copilot code review billing. Review the Copilot code review billing settings.