chore(zenmux): refresh Qwen model family

models/alibaba/qwen-audio-3.0-tts-plus.toml

@@ -0,0 +1,17 @@
+name = "Qwen-Audio-3.0-TTS-Plus"
+description = "Qwen-Audio-3.0-TTS-Plus is Alibaba's higher-quality text-to-speech model, generating spoken audio from text via the DashScope Speech Synthesizer API."
+release_date = "2026-08-05"
+last_updated = "2026-08-05"
+attachment = false
+reasoning = false
+temperature = false
+tool_call = false
+open_weights = false
+
+[limit]
+context = 0
+output = 0
+
+[modalities]
+input = ["text"]
+output = ["audio"]

models/alibaba/qwen-image-2.0-pro.toml

@@ -0,0 +1,18 @@
+# https://www.alibabacloud.com/help/en/model-studio/qwen-image-api
+name = "Qwen-Image-2.0-Pro"
+description = "Qwen-Image-2.0-Pro is an image generation model launched by Qwen AI."
+release_date = "2026-03-03"
+last_updated = "2026-03-03"
+attachment = true
+reasoning = false
+temperature = false
+tool_call = false
+open_weights = false
+
+[limit]
+context = 1_300
+output = 0
+
+[modalities]
+input = ["text", "image"]
+output = ["image"]

models/alibaba/qwen-image-2.0.toml

@@ -0,0 +1,18 @@
+# https://www.alibabacloud.com/help/en/model-studio/qwen-image-api
+name = "Qwen-Image-2.0"
+description = "Qwen-Image-2.0 is an image generation model launched by Qwen AI."
+release_date = "2026-03-03"
+last_updated = "2026-03-03"
+attachment = true
+reasoning = false
+temperature = false
+tool_call = false
+open_weights = false
+
+[limit]
+context = 1_300
+output = 0
+
+[modalities]
+input = ["text", "image"]
+output = ["image"]

models/alibaba/qwen-image-3.0-pro.toml

@@ -0,0 +1,18 @@
+# https://www.alibabacloud.com/help/en/model-studio/qwen-image-generation-and-editing-api-reference
+name = "Qwen-Image-3.0-Pro"
+description = "Rich content: Supports input of up to 4.5k tokens and dense information layout with images-within-images, enabling complex layouts like newspapers, storyboards, menus, and exam papers to be generated in a single pass. Authentic detail: Supports precise rendering of text as small as 10px, and vividly reproduces fine details such as micro-expressions, pores, and individual strands of hair—approaching the quality of real photography. Deep knowledge: Supports native rendering of 12 languages and 20+ fonts, realistic simulation of mainstream interfaces such as web pages, games, and live streams, fully incorporating external knowledge. Qwen-Image-3.0-Pro isn't just pursuing \"good looks\"—it's pursuing \"usefulness\", making image generation a truly deployable productivity tool."
+release_date = "2026-08-05"
+last_updated = "2026-08-05"
+attachment = true
+reasoning = false
+temperature = false
+tool_call = false
+open_weights = false
+
+[limit]
+context = 4_500
+output = 0
+
+[modalities]
+input = ["text", "image"]
+output = ["image"]

models/alibaba/qwen-image-3.0.toml

@@ -0,0 +1,18 @@
+# https://www.alibabacloud.com/help/en/model-studio/qwen-image-generation-and-editing-api-reference
+name = "Qwen-Image-3.0"
+description = "Accurate Prompt Understanding: Supports up to 4.5k token inputs, accurately interpreting complex text-and-image prompts and generating dense layouts in one pass. Reliable Text Rendering: Delivers crisp 10px text across 12 languages and 20+ fonts, making infographics and interfaces ready to use. Efficient Batch Production: Scales everyday tasks like posters, web pages, and UI screens with better cost efficiency, keeping creative output sustainable. Qwen-Image-3.0 Standard is built not just for visual quality, but for smooth, reliable daily creation—turning image generation into sustainable content productivity."
+release_date = "2026-08-05"
+last_updated = "2026-08-05"
+attachment = true
+reasoning = false
+temperature = false
+tool_call = false
+open_weights = false
+
+[limit]
+context = 4_500
+output = 0
+
+[modalities]
+input = ["text", "image"]
+output = ["image"]

models/alibaba/qwen3-14b.toml

@@ -0,0 +1,24 @@
+name = "Qwen3 14B"
+description = "Qwen instruction model for multilingual chat, reasoning, and tool use"
+family = "qwen"
+release_date = "2025-04-28"
+last_updated = "2025-04-28"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2025-04"
+open_weights = true
+license = "Apache 2.0"
+
+[limit]
+context = 131_072
+output = 8_192
+
+[modalities]
+input = ["text"]
+output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/Qwen/Qwen3-14B"

models/alibaba/qwen3-235b-a22b-thinking-2507.toml

@@ -0,0 +1,23 @@
+# https://huggingface.co/Qwen/Qwen3-235B-A22B-Thinking-2507
+name = "Qwen3 235B A22B Thinking 2507"
+description = "Qwen3-235B-A22B-Thinking-2507 is a high-performance, open-weight Mixture-of-Experts (MoE) language model optimized for complex reasoning tasks. It activates 22B of its 235B parameters per forward pass and natively supports up to 262,144 tokens of context. This \"thinking-only\" variant enhances structured logical reasoning, mathematics, science, and long-form generation, showing strong benchmark performance across AIME, SuperGPQA, LiveCodeBench, and MMLU-Redux. It enforces a special reasoning mode (</think>) and is designed for high-token outputs (up to 81,920 tokens) in challenging domains.\n\nThe model is instruction-tuned and excels at step-by-step reasoning, tool use, agentic workflows, and multilingual tasks. This release represents the most capable open-source variant in the Qwen3-235B series, surpassing many closed models in structured reasoning use cases."
+release_date = "2025-07-21"
+last_updated = "2025-07-21"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+open_weights = true
+license = "Apache 2.0"
+
+[limit]
+context = 262_144
+output = 81_920
+
+[modalities]
+input = ["text"]
+output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/Qwen/Qwen3-235B-A22B-Thinking-2507"

models/alibaba/qwen3-asr-flash.toml

@@ -0,0 +1,19 @@
+name = "Qwen3-ASR Flash"
+description = "Speech transcription model for accurate audio-to-text and captioning workflows"
+family = "qwen"
+release_date = "2025-09-08"
+last_updated = "2025-09-08"
+attachment = false
+reasoning = false
+temperature = false
+tool_call = false
+knowledge = "2024-04"
+open_weights = false
+
+[limit]
+context = 53_248
+output = 4_096
+
+[modalities]
+input = ["audio"]
+output = ["text"]

models/alibaba/qwen3-max-2026-01-23.toml

@@ -0,0 +1,25 @@
+# https://docs.modelstudio.console.alibabacloud.com/en/model-studio/model-qwen3-max
+# https://www.alibabacloud.com/blog/pushing-qwen3-max-thinking-beyond-its-limits_602834
+name = "Qwen3 Max 2026-01-23"
+description = "Qwen3 Max snapshot integrating thinking and non-thinking modes for complex reasoning, instruction following, multilingual tasks, retrieval, and tool use"
+family = "qwen"
+release_date = "2026-01-23"
+last_updated = "2026-01-23"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+open_weights = false
+
+[limit]
+context = 262_144
+output = 65_536
+
+[modalities]
+input = ["text"]
+output = ["text"]
+
+[[links]]
+label = "Alibaba Cloud Model Studio"
+url = "https://docs.modelstudio.console.alibabacloud.com/en/model-studio/model-qwen3-max"
+type = "docs"

models/alibaba/qwen3-rerank.toml

@@ -0,0 +1,17 @@
+name = "Qwen3 Rerank"
+description = "Qwen3 Rerank is a new proprietary model of the Qwen model family. Specifically designed for reranking tasks, built on the Qwen3 foundation model. Leveraging Qwen3’s robust multilingual text understanding capabilities, the series achieves state-of-the-art performance across multiple benchmarks for text embedding and reranking tasks."
+release_date = "2026-01-08"
+last_updated = "2026-01-08"
+attachment = false
+reasoning = false
+temperature = false
+tool_call = false
+open_weights = false
+
+[limit]
+context = 8_192
+output = 8_192
+
+[modalities]
+input = ["text"]
+output = ["text"]

models/alibaba/qwen3-vl-embedding.toml

@@ -0,0 +1,17 @@
+name = "Qwen3 VL Embedding"
+description = "The Qwen3-VL-Embedding model series are the latest additions to the Qwen family, built upon the recently open-sourced and powerful Qwen3-VL foundation model. Specifically designed for multimodal information retrieval and cross-modal understanding, this suite accepts diverse inputs including text, images, screenshots, and videos, as well as inputs containing a mixture of these modalities."
+release_date = "2026-01-08"
+last_updated = "2026-01-08"
+attachment = true
+reasoning = false
+temperature = false
+tool_call = false
+open_weights = false
+
+[limit]
+context = 8_192
+output = 8_192
+
+[modalities]
+input = ["text", "video", "image"]
+output = ["text"]

models/alibaba/qwen3-vl-rerank.toml

@@ -0,0 +1,18 @@
+# https://docs.modelstudio.console.alibabacloud.com/en/model-studio/qwen3-vl-rerank
+name = "Qwen3 VL Rerank"
+description = "The Qwen3-VL-Rerank model series are the latest additions to the Qwen family, built upon the recently open-sourced and powerful Qwen3-VL foundation model. Specifically designed for multimodal information retrieval and cross-modal understanding, this suite accepts diverse inputs including text, images, screenshots, and videos, as well as inputs containing a mixture of these modalities."
+release_date = "2026-01-08"
+last_updated = "2026-01-08"
+attachment = true
+reasoning = false
+temperature = false
+tool_call = false
+open_weights = false
+
+[limit]
+context = 120_000
+output = 0
+
+[modalities]
+input = ["text", "image", "video"]
+output = ["text"]

providers/zenmux/models/qwen/qwen-audio-3.0-tts-plus.toml

@@ -0,0 +1,3 @@
+# ZenMux pricing is non-token-denominated for this route and is intentionally omitted.
+# https://zenmux.ai/docs/api/openai/openai-list-models.html
+base_model = "alibaba/qwen-audio-3.0-tts-plus"

providers/zenmux/models/qwen/qwen-image-2.0-pro.toml

@@ -0,0 +1,7 @@
+# ZenMux pricing is non-token-denominated for this route and is intentionally omitted.
+# Host prompt limit: the 2026-09-19 ZenMux page reports context_length=1024.
+# https://zenmux.ai/docs/api/openai/openai-list-models.html
+base_model = "alibaba/qwen-image-2.0-pro"
+
+[limit]
+context = 1_024

providers/zenmux/models/qwen/qwen-image-2.0.toml

@@ -0,0 +1,7 @@
+# ZenMux pricing is non-token-denominated for this route and is intentionally omitted.
+# Host prompt limit: the 2026-09-19 ZenMux page reports context_length=1024.
+# https://zenmux.ai/docs/api/openai/openai-list-models.html
+base_model = "alibaba/qwen-image-2.0"
+
+[limit]
+context = 1_024

providers/zenmux/models/qwen/qwen-image-3.0-pro.toml

@@ -0,0 +1,7 @@
+# ZenMux pricing is non-token-denominated for this route and is intentionally omitted.
+# Host prompt limit: the 2026-09-19 ZenMux page reports context_length=1024.
+# https://zenmux.ai/docs/api/openai/openai-list-models.html
+base_model = "alibaba/qwen-image-3.0-pro"
+
+[limit]
+context = 1_024

providers/zenmux/models/qwen/qwen-image-3.0.toml

@@ -0,0 +1,7 @@
+# ZenMux pricing is non-token-denominated for this route and is intentionally omitted.
+# Host prompt limit: the 2026-09-19 ZenMux page reports context_length=1024.
+# https://zenmux.ai/docs/api/openai/openai-list-models.html
+base_model = "alibaba/qwen-image-3.0"
+
+[limit]
+context = 1_024

providers/zenmux/models/qwen/qwen3-14b.toml

@@ -0,0 +1,21 @@
+# Toggle: reasoning.enabled = true|false
+# Budget: reasoning.max_tokens (integer reasoning tokens)
+# https://zenmux.ai/docs/api/openai/openai-list-models.html
+# https://zenmux.ai/docs/guide/advanced/reasoning.html
+base_model = "alibaba/qwen3-14b"
+name = "Qwen3-14B"
+
+[[reasoning_options]]
+type = "toggle"
+
+[[reasoning_options]]
+type = "budget_tokens"
+
+[cost]
+input = 0.14
+output = 1.4
+cache_read = 0.028
+
+[limit]
+context = 32_000
+output = 32_000

providers/zenmux/models/qwen/qwen3-235b-a22b-2507.toml

@@ -0,0 +1,14 @@
+# Host limits: the 2026-09-19 ZenMux page reports context_length=256000 and max_completion_tokens=128000.
+# https://zenmux.ai/docs/api/openai/openai-list-models.html
+base_model = "alibaba/qwen3-235b-a22b-instruct-2507"
+name = "Qwen3 235B A22B Instruct 2507"
+structured_output = true
+
+[cost]
+input = 0.28
+output = 1.11
+cache_read = 0.056
+
+[limit]
+context = 256_000
+output = 128_000

providers/zenmux/models/qwen/qwen3-235b-a22b-thinking-2507.toml

@@ -0,0 +1,18 @@
+# Host limits: the 2026-09-19 ZenMux page reports context_length=256000 and max_completion_tokens=128000.
+# https://zenmux.ai/docs/api/openai/openai-list-models.html
+# https://zenmux.ai/docs/guide/advanced/reasoning.html
+base_model = "alibaba/qwen3-235b-a22b-thinking-2507"
+structured_output = true
+reasoning_options = []
+
+[interleaved]
+field = "reasoning_content"
+
+[cost]
+input = 0.28
+output = 2.78
+cache_read = 0.056
+
+[limit]
+context = 256_000
+output = 128_000

providers/zenmux/models/qwen/qwen3-asr-flash.toml

@@ -0,0 +1,13 @@
+# ZenMux's list flags this route as reasoning-capable, but the transcription API
+# exposes neither a reasoning request control nor a reasoning response channel;
+# lab and first-party metadata both classify Qwen3 ASR Flash as non-reasoning.
+# Host limits: page max_context=1000000 and max_completion_tokens=65536;
+# OpenAI list independently reports context_length=1000000.
+# ZenMux pricing is non-token-denominated for this route and is intentionally omitted.
+# https://zenmux.ai/docs/api/openai/openai-list-models.html
+base_model = "alibaba/qwen3-asr-flash"
+name = "Qwen3 ASR Flash"
+
+[limit]
+context = 1_000_000
+output = 65_536