Introduction: the multimodal Thinking edition
Qwen3-VL-235B-A22B-Thinking is the reasoning-focused mixture-of-experts edition of Qwen3-VL. It combines text with images and video, producing textual answers or code. The fixed Thinking checkpoint is separate from Qwen3-VL-235B-A22B-Instruct and the separately named FP8 builds. Choose it for a question that requires connecting visual evidence with a longer explanation, rather than assuming every Qwen3-VL edition has the same response style.
Visual agents and visual coding
Qwen describes a visual-agent workflow that recognizes elements in computer or mobile interfaces, interprets their functions and works with tools to complete tasks. The model supplies perception and reasoning; an application must connect and execute the actions. For an interface review, provide the current screenshot and a specific goal, then ask it to identify the relevant controls and explain the next action. The image alone does not confer control of the device.
Visual coding includes generating diagram content and HTML, CSS or JavaScript from images and videos. Supply a screenshot with the required layout, behavior and editable output format. Ask for the page structure and styles as separate artifacts when that makes review easier. Generated code is text for a development workflow; it is not evidence that a website has been rendered or that the source image has become a working application.
Spatial perception
The maker documents reasoning about object positions, viewpoints and occlusions, with two-dimensional grounding and support for three-dimensional grounding. Ask a concrete relationship question and identify the object of interest: for example, which component is obscured, what is in front of it, and which visual clues support that interpretation. Grounding capability should not be treated as a calibrated physical measurement from an arbitrary picture.
Long context and video
The native context is 262,144 tokens. Qwen describes expansion to roughly one million through YaRN configuration, and family workflows involving books or hours-long video with event indexing at second-level granularity. These capabilities depend on representation, frame sampling and context configuration; the broad maker description does not guarantee perfect recall of every frame. In its repository, Qwen advises an extrapolation factor of two or three for this interleaved positional scheme when extending 256K to about 1M, rather than automatically copying a factor of four from another model.
Multimodal reasoning and recognition
The Thinking edition targets evidence-based STEM and mathematical reasoning, causal analysis and logical answers that connect text with visual material. Give a diagram and its definitions together, or a chart and the precise numerical question. Request the labels and values used before the conclusion. Qwen also describes broader visual recognition across people, anime, products, landmarks, plants and animals. A recognition answer remains an inference from the supplied image, not a guarantee that every object or identity can be named.
OCR covers 32 languages and is designed to improve reading under low light, blur or tilt, including specialized vocabulary and long-document structure. Preserve legible source images and ask for fields, quotations or a structured table that can be checked. Text and visual understanding are combined in one model, so a question can refer jointly to a document passage, a figure and the user’s instructions.
Architecture changes
Interleaved-MRoPE distributes positional representation across time, width and height for video reasoning. DeepStack combines visual features from several encoder levels to capture finer details and improve image–text alignment. Text–timestamp alignment grounds events in temporal references. The released text configuration has 94 layers, 128 experts and eight experts selected per token. These architectural details explain the model’s multimodal representation; they do not establish a response-time guarantee.
Preparing visual input and output
The maker’s examples cover single or multiple images, video and text-only requests. Label each visual input so the question can refer to it clearly. Image and video processors use separate pixel budgets: an image budget limits its pixels, while a video budget accounts for pixels across the sampled frames. Explicit image dimensions are aligned to multiples of 32. Video sampling can be controlled by rate or frame count; the documented default is two frames per second. Preserve timestamps when asking for an event timeline.
For a long clip, ask for the event, its time interval and the visible evidence rather than an undifferentiated summary. For several pages, request a result organized by page identifier. These preparation examples apply the documented capabilities and help make omissions or extraction errors visible.
Generation guidance and limits
The checkpoint’s generation file defaults to temperature 0.8, top_p 0.95 and top_k 20. Qwen’s separate Thinking evaluation recipe uses temperature 0.6, presence penalty zero, repetition penalty 1.0 and a 40,960-token output allowance. These are distinct producer presets, not a single hard output cap. Allow space for reasoning and final text alongside the visual tokens, and verify extracted text, calculated answers and generated code against the source before using them.
", "aboutText": "Introduction: the multimodal Thinking edition\n\nQwen3-VL-235B-A22B-Thinking is the reasoning-focused mixture-of-experts edition of Qwen3-VL. It combines text with images and video, producing textual answers or code. The fixed Thinking checkpoint is separate from Qwen3-VL-235B-A22B-Instruct and the separately named FP8 builds. Choose it for a question that requires connecting visual evidence with a longer explanation, rather than assuming every Qwen3-VL edition has the same response style.\n\nVisual agents and visual coding\n\nQwen describes a visual-agent workflow that recognizes elements in computer or mobile interfaces, interprets their functions and works with tools to complete tasks. The model supplies perception and reasoning; an application must connect and execute the actions. For an interface review, provide the current screenshot and a specific goal, then ask it to identify the relevant controls and explain the next action. The image alone does not confer control of the device.\n\nVisual coding includes generating diagram content and HTML, CSS or JavaScript from images and videos. Supply a screenshot with the required layout, behavior and editable output format. Ask for the page structure and styles as separate artifacts when that makes review easier. Generated code is text for a development workflow; it is not evidence that a website has been rendered or that the source image has become a working application.\n\nSpatial perception\n\nThe maker documents reasoning about object positions, viewpoints and occlusions, with two-dimensional grounding and support for three-dimensional grounding. Ask a concrete relationship question and identify the object of interest: for example, which component is obscured, what is in front of it, and which visual clues support that interpretation. Grounding capability should not be treated as a calibrated physical measurement from an arbitrary picture.\n\nLong context and video\n\nThe native context is 262,144 tokens. Qwen describes expansion to roughly one million through YaRN configuration, and family workflows involving books or hours-long video with event indexing at second-level granularity. These capabilities depend on representation, frame sampling and context configuration; the broad maker description does not guarantee perfect recall of every frame. In its repository, Qwen advises an extrapolation factor of two or three for this interleaved positional scheme when extending 256K to about 1M, rather than automatically copying a factor of four from another model.\n\nMultimodal reasoning and recognition\n\nThe Thinking edition targets evidence-based STEM and mathematical reasoning, causal analysis and logical answers that connect text with visual material. Give a diagram and its definitions together, or a chart and the precise numerical question. Request the labels and values used before the conclusion. Qwen also describes broader visual recognition across people, anime, products, landmarks, plants and animals. A recognition answer remains an inference from the supplied image, not a guarantee that every object or identity can be named.\n\nOCR covers 32 languages and is designed to improve reading under low light, blur or tilt, including specialized vocabulary and long-document structure. Preserve legible source images and ask for fields, quotations or a structured table that can be checked. Text and visual understanding are combined in one model, so a question can refer jointly to a document passage, a figure and the user’s instructions.\n\nArchitecture changes\n\nInterleaved-MRoPE distributes positional representation across time, width and height for video reasoning. DeepStack combines visual features from several encoder levels to capture finer details and improve image–text alignment. Text–timestamp alignment grounds events in temporal references. The released text configuration has 94 layers, 128 experts and eight experts selected per token. These architectural details explain the model’s multimodal representation; they do not establish a response-time guarantee.\n\nPreparing visual input and output\n\nThe maker’s examples cover single or multiple images, video and text-only requests. Label each visual input so the question can refer to it clearly. Image and video processors use separate pixel budgets: an image budget limits its pixels, while a video budget accounts for pixels across the sampled frames. Explicit image dimensions are aligned to multiples of 32. Video sampling can be controlled by rate or frame count; the documented default is two frames per second. Preserve timestamps when asking for an event timeline.\n\nFor a long clip, ask for the event, its time interval and the visible evidence rather than an undifferentiated summary. For several pages, request a result organized by page identifier. These preparation examples apply the documented capabilities and help make omissions or extraction errors visible.\n\nGeneration guidance and limits\n\nThe checkpoint’s generation file defaults to temperature 0.8, top_p 0.95 and top_k 20. Qwen’s separate Thinking evaluation recipe uses temperature 0.6, presence penalty zero, repetition penalty 1.0 and a 40,960-token output allowance. These are distinct producer presets, not a single hard output cap. Allow space for reasoning and final text alongside the visual tokens, and verify extracted text, calculated answers and generated code against the source before using them.", "editorialOverlay": { "copy_fields": { "practical_html": "Technical example in English
Reason from a textual circuit description\n\nUse this textual circuit description: an ideal 12 V source powers two series resistors, R1 = 2 Ω and R2 = 4 Ω. Calculate total resistance, current and voltage across each resistor. State the ideal-component assumptions and verify that the voltage drops sum to the source voltage. Return a compact worked solution using the labels R1 and R2.A labeled worked calculation and a voltage-sum consistency check.The same series current is used for both resistors and the stated voltage drops sum to 12 V.qwen/qwen3-vl-235b-a22b-thinkingQwen3-VL-235B-A22B-Thinking is Qwen’s mixture-of-experts vision-language model for text, image and video reasoning. Use the Thinking edition for STEM questions, document evidence, spatial relationships, visual coding and timestamped video analysis, with a native 262,144-token context.
This description covers the model’s capabilities. Available in Ailoot inputs, outputs and options are listed for the selected method below.
Introduction: the multimodal Thinking edition
Qwen3-VL-235B-A22B-Thinking is the reasoning-focused mixture-of-experts edition of Qwen3-VL. It combines text with images and video, producing textual answers or code. The fixed Thinking checkpoint is separate from Qwen3-VL-235B-A22B-Instruct and the separately named FP8 builds. Choose it for a question that requires connecting visual evidence with a longer explanation, rather than assuming every Qwen3-VL edition has the same response style.
Visual agents and visual coding
Qwen describes a visual-agent workflow that recognizes elements in computer or mobile interfaces, interprets their functions and works with tools to complete tasks. The model supplies perception and reasoning; an application must connect and execute the actions. For an interface review, provide the current screenshot and a specific goal, then ask it to identify the relevant controls and explain the next action. The image alone does not confer control of the device.
Visual coding includes generating diagram content and HTML, CSS or JavaScript from images and videos. Supply a screenshot with the required layout, behavior and editable output format. Ask for the page structure and styles as separate artifacts when that makes review easier. Generated code is text for a development workflow; it is not evidence that a website has been rendered or that the source image has become a working application.
Spatial perception
The maker documents reasoning about object positions, viewpoints and occlusions, with two-dimensional grounding and support for three-dimensional grounding. Ask a concrete relationship question and identify the object of interest: for example, which component is obscured, what is in front of it, and which visual clues support that interpretation. Grounding capability should not be treated as a calibrated physical measurement from an arbitrary picture.
Long context and video
The native context is 262,144 tokens. Qwen describes expansion to roughly one million through YaRN configuration, and family workflows involving books or hours-long video with event indexing at second-level granularity. These capabilities depend on representation, frame sampling and context configuration; the broad maker description does not guarantee perfect recall of every frame. In its repository, Qwen advises an extrapolation factor of two or three for this interleaved positional scheme when extending 256K to about 1M, rather than automatically copying a factor of four from another model.
Multimodal reasoning and recognition
The Thinking edition targets evidence-based STEM and mathematical reasoning, causal analysis and logical answers that connect text with visual material. Give a diagram and its definitions together, or a chart and the precise numerical question. Request the labels and values used before the conclusion. Qwen also describes broader visual recognition across people, anime, products, landmarks, plants and animals. A recognition answer remains an inference from the supplied image, not a guarantee that every object or identity can be named.
OCR covers 32 languages and is designed to improve reading under low light, blur or tilt, including specialized vocabulary and long-document structure. Preserve legible source images and ask for fields, quotations or a structured table that can be checked. Text and visual understanding are combined in one model, so a question can refer jointly to a document passage, a figure and the user’s instructions.
Architecture changes
Interleaved-MRoPE distributes positional representation across time, width and height for video reasoning. DeepStack combines visual features from several encoder levels to capture finer details and improve image–text alignment. Text–timestamp alignment grounds events in temporal references. The released text configuration has 94 layers, 128 experts and eight experts selected per token. These architectural details explain the model’s multimodal representation; they do not establish a response-time guarantee.
Preparing visual input and output
The maker’s examples cover single or multiple images, video and text-only requests. Label each visual input so the question can refer to it clearly. Image and video processors use separate pixel budgets: an image budget limits its pixels, while a video budget accounts for pixels across the sampled frames. Explicit image dimensions are aligned to multiples of 32. Video sampling can be controlled by rate or frame count; the documented default is two frames per second. Preserve timestamps when asking for an event timeline.
For a long clip, ask for the event, its time interval and the visible evidence rather than an undifferentiated summary. For several pages, request a result organized by page identifier. These preparation examples apply the documented capabilities and help make omissions or extraction errors visible.
Generation guidance and limits
The checkpoint’s generation file defaults to temperature 0.8, top_p 0.95 and top_k 20. Qwen’s separate Thinking evaluation recipe uses temperature 0.6, presence penalty zero, repetition penalty 1.0 and a 40,960-token output allowance. These are distinct producer presets, not a single hard output cap. Allow space for reasoning and final text alongside the visual tokens, and verify extracted text, calculated answers and generated code against the source before using them.
POST /v1/chat/completionstexttext| Field | Type | Requirement | Values & limits | Default |
|---|---|---|---|---|
Fieldmodel | Typestring | RequirementRequired | Values & limits qwen/qwen3-vl-235b-a22b-thinking | Default— |
Fieldmessages | Typemessage[] | RequirementRequired | Values & limits Minimum items: 1 | Default— |
Fieldmessages[].role | Typestring | RequirementRequired | Values & limits Model-specific | Default— |
Fieldmessages[].content | Typestring | RequirementOptional | Values & limits Applies when: Text conversation | Default— |
Fieldstream | Typeboolean | RequirementOptional | Values & limits | Defaultfalse |
Text · UTF8 text
messages[].content
| Parameter | Contract |
|---|---|
Parametercontent_type | Contractapplication/json |
Parameterrequired | Contractmodel, messages |
Parametermaterial_types | Contracttext |
Parametermessages | Contractmessage array with role/content shape |
Parameterfile_material | Contractactual UTF-8 document inserted as literal text; no file ID inferred |
Parameteroptional_controls | Contractstream |
Parametercontrol_scope | ContractText conversation; stream is a boolean and defaults to false. |
Parameteroutput_modalities | Contracttext |
Parameterhosted_operations | Contractfalse |
Parametermultimodal_input_binding | Contractnull |
Method aliases: /chat/completions
Lifecycle: synchronous_or_streamed
Send exact model ID with messages.
Context length is not an output cap.
A text file may supply literal text; image/audio/file-object inputs need exact binding.
Tariff references do not add request options. Supported parameters are listed for the selected method.
| Configuration | Rate schedule | Charge |
|---|---|---|
| ConfigurationBase token rates | Rate scheduleBase rate | Charge Input · Input tokens $0.000572 / 1K tokens Output · Output tokens $0.00572 / 1K tokens Input · Cache read $0 / 1K tokens Input · Cache write $0 / 1K tokens Input · Cache write, 1 hour — / 1K tokens |
Ailoot reserves an estimated amount before a request. After the result, it settles the final charge from measured usage. Streaming settlement uses the final usage record. How reservation and settlement work →
set -eu\n: "${AILOOT_API_KEY:?Set your Ailoot API key}"\npython3 - <<'PY_REQUEST'\nimport json,os\nfrom pathlib import Path\np=json.loads('{"model":"qwen/qwen3-vl-235b-a22b-thinking","messages":[{"role":"user","content":"hello"}],"stream":false}')\n\nPath("request.json").write_text(json.dumps(p,ensure_ascii=False))\nPY_REQUEST\ncurl --fail-with-body --silent --show-error https://api.ailoot.ai/v1/chat/completions \\\n -H "Authorization: Bearer $AILOOT_API_KEY" -H 'Content-Type: application/json' \\\n --data-binary @request.json -D response-headers.txt -o result.json\npython3 - <<'PY_RESULT'\nimport json\nfrom pathlib import Path\nr=json.loads(Path("result.json").read_text())\nfor x in r.get("choices",[]):\n if x.get("message",{}).get("content"): print(x["message"]["content"])\nPY_RESULT\nfrom openai import OpenAI\nimport os\n\nif not os.environ.get("AILOOT_API_KEY", "").strip(): raise SystemExit("Set AILOOT_API_KEY before making a request")\nclient=OpenAI(base_url="https://api.ailoot.ai/v1",api_key=os.environ["AILOOT_API_KEY"],max_retries=0)\nresult=client.chat.completions.create(model="qwen/qwen3-vl-235b-a22b-thinking", messages=[{"role":"user","content":"hello"}], stream=False)\nimport base64, json\nfrom pathlib import Path\n\ndef save_result(data, stem="result"):\n # Keep every field, including usage, IDs, reasoning and tool requests.\n Path(stem + ".json").write_text(json.dumps(data, ensure_ascii=False, indent=2), encoding="utf-8")\n print(json.dumps(data, ensure_ascii=False, indent=2))\n def save_blob(encoded, suffix, label):\n raw = base64.b64decode(encoded, altchars=b"-_", validate=True)\n ext = "png" if raw.startswith(b"\\x89PNG\\r\\n\\x1a\\n") else "jpg" if raw.startswith(b"\\xff\\xd8") else "webp" if raw[:4] == b"RIFF" and raw[8:12] == b"WEBP" else suffix\n file = Path(stem + "-" + label + "." + ext)\n file.write_bytes(raw)\n print("Saved", file)\n for index, block in enumerate(data.get("content") or []):\n print("Content", index, json.dumps(block, ensure_ascii=False))\n for choice in data.get("choices") or []:\n print("Choice", choice.get("index"), json.dumps(choice.get("message", choice.get("delta", {})), ensure_ascii=False))\n for index, item in enumerate(data.get("output") or []):\n print("Output", index, json.dumps(item, ensure_ascii=False))\n if item.get("type") == "image_generation_call" and item.get("result"):\n save_blob(item["result"], item.get("output_format") or "bin", "output-" + str(index))\n for ci, candidate in enumerate(data.get("candidates") or []):\n for pi, part in enumerate((candidate.get("content") or {}).get("parts") or []):\n print("Candidate/part", ci, pi, json.dumps(part, ensure_ascii=False))\n blob = part.get("inlineData") or part.get("inline_data")\n if blob and blob.get("data"):\n mime = blob.get("mimeType") or blob.get("mime_type")\n suffix = {"image/png":"png", "image/jpeg":"jpg", "image/webp":"webp", "audio/wav":"wav", "audio/mpeg":"mp3", "video/mp4":"mp4"}.get(mime, "bin")\n save_blob(blob["data"], suffix, "candidate-" + str(ci) + "-part-" + str(pi))\n\nsave_result(result.to_dict(mode="json"))\njson or SSE
Keep every complete returned response, file and content part.
Keep every choice and its content/tool calls/reasoning. In SSE assemble deltas by choice/tool index, retain the final finish reason and usage.
Keep request/response IDs and headers. No generic text job retrieval URL is supplied.
| Field | Type |
|---|---|
Fieldchoices[].index | Typeinteger |
Fieldchoices[].message.content | Typestring|null |
Fieldchoices[].message.tool_calls | Typetool_call[] |
Fieldchoices[].message.reasoning_content | Typestring |
Fieldchoices[].message.reasoning_details | Typeobject|array |
Fieldchoices[].finish_reason | Typestring|null |
Fieldusage | Typeobject |
| Response | Condition | Next step |
|---|---|---|
Response400 · invalid_request_error | ConditionInvalid required fields, material representation or parameter value | Next stepCorrect the indicated fields before a new creation request. |
Ailoot Ultra qwen/qwen3-vl-235b-a22b-thinking, qwen3-vl-235b-a22b-thinking
/chat/completions
qwen/qwen3-vl-235b-a22b-thinking, qwen3-vl-235b-a22b-thinking
qwen/qwen3-vl-235b-a22b-thinkingQwen3-VL-235B-A22B-Thinking is Qwen’s mixture-of-experts vision-language model for text, image and video reasoning. Use the Thinking edition for STEM questions, document evidence, spatial relationships, visual coding and timestamped video analysis, with a native 262,144-token context.
This description covers the model’s capabilities. Available in Ailoot inputs, outputs and options are listed for the selected method below.
Introduction: the multimodal Thinking edition
Qwen3-VL-235B-A22B-Thinking is the reasoning-focused mixture-of-experts edition of Qwen3-VL. It combines text with images and video, producing textual answers or code. The fixed Thinking checkpoint is separate from Qwen3-VL-235B-A22B-Instruct and the separately named FP8 builds. Choose it for a question that requires connecting visual evidence with a longer explanation, rather than assuming every Qwen3-VL edition has the same response style.
Visual agents and visual coding
Qwen describes a visual-agent workflow that recognizes elements in computer or mobile interfaces, interprets their functions and works with tools to complete tasks. The model supplies perception and reasoning; an application must connect and execute the actions. For an interface review, provide the current screenshot and a specific goal, then ask it to identify the relevant controls and explain the next action. The image alone does not confer control of the device.
Visual coding includes generating diagram content and HTML, CSS or JavaScript from images and videos. Supply a screenshot with the required layout, behavior and editable output format. Ask for the page structure and styles as separate artifacts when that makes review easier. Generated code is text for a development workflow; it is not evidence that a website has been rendered or that the source image has become a working application.
Spatial perception
The maker documents reasoning about object positions, viewpoints and occlusions, with two-dimensional grounding and support for three-dimensional grounding. Ask a concrete relationship question and identify the object of interest: for example, which component is obscured, what is in front of it, and which visual clues support that interpretation. Grounding capability should not be treated as a calibrated physical measurement from an arbitrary picture.
Long context and video
The native context is 262,144 tokens. Qwen describes expansion to roughly one million through YaRN configuration, and family workflows involving books or hours-long video with event indexing at second-level granularity. These capabilities depend on representation, frame sampling and context configuration; the broad maker description does not guarantee perfect recall of every frame. In its repository, Qwen advises an extrapolation factor of two or three for this interleaved positional scheme when extending 256K to about 1M, rather than automatically copying a factor of four from another model.
Multimodal reasoning and recognition
The Thinking edition targets evidence-based STEM and mathematical reasoning, causal analysis and logical answers that connect text with visual material. Give a diagram and its definitions together, or a chart and the precise numerical question. Request the labels and values used before the conclusion. Qwen also describes broader visual recognition across people, anime, products, landmarks, plants and animals. A recognition answer remains an inference from the supplied image, not a guarantee that every object or identity can be named.
OCR covers 32 languages and is designed to improve reading under low light, blur or tilt, including specialized vocabulary and long-document structure. Preserve legible source images and ask for fields, quotations or a structured table that can be checked. Text and visual understanding are combined in one model, so a question can refer jointly to a document passage, a figure and the user’s instructions.
Architecture changes
Interleaved-MRoPE distributes positional representation across time, width and height for video reasoning. DeepStack combines visual features from several encoder levels to capture finer details and improve image–text alignment. Text–timestamp alignment grounds events in temporal references. The released text configuration has 94 layers, 128 experts and eight experts selected per token. These architectural details explain the model’s multimodal representation; they do not establish a response-time guarantee.
Preparing visual input and output
The maker’s examples cover single or multiple images, video and text-only requests. Label each visual input so the question can refer to it clearly. Image and video processors use separate pixel budgets: an image budget limits its pixels, while a video budget accounts for pixels across the sampled frames. Explicit image dimensions are aligned to multiples of 32. Video sampling can be controlled by rate or frame count; the documented default is two frames per second. Preserve timestamps when asking for an event timeline.
For a long clip, ask for the event, its time interval and the visible evidence rather than an undifferentiated summary. For several pages, request a result organized by page identifier. These preparation examples apply the documented capabilities and help make omissions or extraction errors visible.
Generation guidance and limits
The checkpoint’s generation file defaults to temperature 0.8, top_p 0.95 and top_k 20. Qwen’s separate Thinking evaluation recipe uses temperature 0.6, presence penalty zero, repetition penalty 1.0 and a 40,960-token output allowance. These are distinct producer presets, not a single hard output cap. Allow space for reasoning and final text alongside the visual tokens, and verify extracted text, calculated answers and generated code against the source before using them.
POST /v1/messagestextjson, sse| Field | Type | Requirement | Values & limits | Default |
|---|---|---|---|---|
Fieldmodel | Typestring | RequirementRequired | Values & limits qwen/qwen3-vl-235b-a22b-thinking | Default— |
Fieldmessages | Typemessage[] | RequirementRequired | Values & limits | Default— |
Fieldmessages[].role | Typestring | RequirementRequired | Values & limits user, assistant | Default— |
Fieldmessages[].content | Typestring | RequirementRequired | Values & limits | Default— |
Fieldmax_tokens | Typeinteger | RequirementRequired | Values & limits | Default— |
Fieldstream | Typeboolean | RequirementOptional | Values & limits | Defaultfalse |
Text · UTF-8 text
messages[].content string
| Parameter | Contract |
|---|---|
Parametercontent_type | Contractapplication/json |
Parameterrequired | Contractmodel, messages, max_tokens |
Parameteroptional_controls | Contractstream |
Parametermaterial_types | Contracttext |
Parametertext_file_material | ContractRead the actual UTF-8 document and put its text in the indicated input or message content string. |
Lifecycle: synchronous_or_streamed
Tariff references do not add request options. Supported parameters are listed for the selected method.
| Configuration | Rate schedule | Charge |
|---|---|---|
| ConfigurationBase token rates | Rate scheduleBase rate | Charge Input · Input tokens $0.000572 / 1K tokens Output · Output tokens $0.00572 / 1K tokens Input · Cache read $0 / 1K tokens Input · Cache write $0 / 1K tokens Input · Cache write, 1 hour — / 1K tokens |
Ailoot reserves an estimated amount before a request. After the result, it settles the final charge from measured usage. Streaming settlement uses the final usage record. How reservation and settlement work →
{\n "model": "qwen/qwen3-vl-235b-a22b-thinking",\n "max_tokens": 128,\n "messages": [\n {\n "role": "user",\n "content": "Summarize the incident report."\n }\n ],\n "stream": false\n}set -eu\n: "${AILOOT_API_KEY:?Set your Ailoot API key}"\n: "${AILOOT_TEXT_FILE:?Set your actual UTF-8 document}"\nexport AILOOT_TEXT_FILE\npython3 - <<'PY_REQUEST'\nimport json, os\nfrom pathlib import Path\n\nsource_file = Path(os.environ['AILOOT_TEXT_FILE'])\nif not source_file.is_file():\n raise SystemExit('Set AILOOT_TEXT_FILE to your actual UTF-8 document')\ntext = source_file.read_text(encoding='utf-8')\nif not text.strip():\n raise SystemExit('Choose a document containing text')\npayload = {'model': 'qwen/qwen3-vl-235b-a22b-thinking', 'max_tokens': 128, 'messages': [{'role': 'user', 'content': 'Summarize the incident report.'}]}\npayload['messages'][0]['content'] = 'Summarize this material and quote supporting source passages.\\n\\n' + text\nPath('request.json').write_text(json.dumps(payload, ensure_ascii=False), encoding='utf-8')\nPY_REQUEST\ncurl --fail-with-body --silent --show-error https://api.ailoot.ai/v1/messages \\\n -H "x-api-key: $AILOOT_API_KEY" -H 'anthropic-version: 2023-06-01' -H 'Content-Type: application/json' \\\n --data-binary @request.json -D response-headers.txt -o result.json\npython3 - <<'PY_RESULT'\nimport json\nfrom pathlib import Path\n\nresult = json.loads(Path('result.json').read_text(encoding='utf-8'))\nprint(json.dumps(result, ensure_ascii=False, indent=2))\nPY_RESULT\nimport json, os\nfrom pathlib import Path\n\nsource_file = Path(os.environ['AILOOT_TEXT_FILE'])\nif not source_file.is_file():\n raise SystemExit('Set AILOOT_TEXT_FILE to your actual UTF-8 document')\ntext = source_file.read_text(encoding='utf-8')\nif not text.strip():\n raise SystemExit('Choose a document containing text')\npayload = {'model': 'qwen/qwen3-vl-235b-a22b-thinking', 'max_tokens': 128, 'messages': [{'role': 'user', 'content': 'Summarize the incident report.'}]}\npayload['messages'][0]['content'] = 'Summarize this material and quote supporting source passages.\\n\\n' + text\nfrom urllib.request import Request, urlopen\nfrom urllib.error import HTTPError\n\napi_key = os.environ['AILOOT_API_KEY']\nif not api_key.strip():\n raise SystemExit('Set AILOOT_API_KEY before making a request')\nheaders = {'x-api-key': 'PLACEHOLDER_KEY', 'anthropic-version': '2023-06-01', 'Content-Type': 'application/json'}\nheaders['x-api-key'] = api_key\nrequest = Request('https://api.ailoot.ai/v1/messages', data=json.dumps(payload, ensure_ascii=False).encode('utf-8'), headers=headers, method='POST')\ntry:\n with urlopen(request, timeout=120) as response:\n raw = response.read()\n Path('response-headers.txt').write_text(str(response.headers), encoding='utf-8')\nexcept HTTPError as error:\n Path('result.json').write_bytes(error.read())\n Path('response-headers.txt').write_text(str(error.headers), encoding='utf-8')\n raise\nPath('result.json').write_bytes(raw)\nimport json\nfrom pathlib import Path\n\nresult = json.loads(Path('result.json').read_text(encoding='utf-8'))\nprint(json.dumps(result, ensure_ascii=False, indent=2))\nJSON or SSE
Keep every complete returned response, file and content part.
Keep all content blocks, thinking signatures and actual tool-use IDs. Assemble all SSE block deltas; a first text block is not the complete output.
Keep request/response IDs and headers. No generic text job retrieval URL is supplied.
| Field | Type |
|---|---|
Fieldid | Typestring |
Fieldmodel | Typestring |
Fieldcontent[] | Typetext|thinking|tool_use|other content block |
Fieldstop_reason | Typestring|null |
Fieldusage | Typeobject |
| Response | Condition | Next step |
|---|---|---|
Response400 · invalid_request_error | ConditionInvalid required fields, material representation or parameter value | Next stepCorrect the indicated fields before a new creation request. |
Ailoot Ultra qwen/qwen3-vl-235b-a22b-thinking, qwen3-vl-235b-a22b-thinking
/chat/completions
qwen/qwen3-vl-235b-a22b-thinking, qwen3-vl-235b-a22b-thinking
qwen3-vl-235b-a22b-thinkingQwen3-VL-235B-A22B-Thinking is Qwen’s mixture-of-experts vision-language model for text, image and video reasoning. Use the Thinking edition for STEM questions, document evidence, spatial relationships, visual coding and timestamped video analysis, with a native 262,144-token context.
This description covers the model’s capabilities. Available in Ailoot inputs, outputs and options are listed for the selected method below.
Introduction: the multimodal Thinking edition
Qwen3-VL-235B-A22B-Thinking is the reasoning-focused mixture-of-experts edition of Qwen3-VL. It combines text with images and video, producing textual answers or code. The fixed Thinking checkpoint is separate from Qwen3-VL-235B-A22B-Instruct and the separately named FP8 builds. Choose it for a question that requires connecting visual evidence with a longer explanation, rather than assuming every Qwen3-VL edition has the same response style.
Visual agents and visual coding
Qwen describes a visual-agent workflow that recognizes elements in computer or mobile interfaces, interprets their functions and works with tools to complete tasks. The model supplies perception and reasoning; an application must connect and execute the actions. For an interface review, provide the current screenshot and a specific goal, then ask it to identify the relevant controls and explain the next action. The image alone does not confer control of the device.
Visual coding includes generating diagram content and HTML, CSS or JavaScript from images and videos. Supply a screenshot with the required layout, behavior and editable output format. Ask for the page structure and styles as separate artifacts when that makes review easier. Generated code is text for a development workflow; it is not evidence that a website has been rendered or that the source image has become a working application.
Spatial perception
The maker documents reasoning about object positions, viewpoints and occlusions, with two-dimensional grounding and support for three-dimensional grounding. Ask a concrete relationship question and identify the object of interest: for example, which component is obscured, what is in front of it, and which visual clues support that interpretation. Grounding capability should not be treated as a calibrated physical measurement from an arbitrary picture.
Long context and video
The native context is 262,144 tokens. Qwen describes expansion to roughly one million through YaRN configuration, and family workflows involving books or hours-long video with event indexing at second-level granularity. These capabilities depend on representation, frame sampling and context configuration; the broad maker description does not guarantee perfect recall of every frame. In its repository, Qwen advises an extrapolation factor of two or three for this interleaved positional scheme when extending 256K to about 1M, rather than automatically copying a factor of four from another model.
Multimodal reasoning and recognition
The Thinking edition targets evidence-based STEM and mathematical reasoning, causal analysis and logical answers that connect text with visual material. Give a diagram and its definitions together, or a chart and the precise numerical question. Request the labels and values used before the conclusion. Qwen also describes broader visual recognition across people, anime, products, landmarks, plants and animals. A recognition answer remains an inference from the supplied image, not a guarantee that every object or identity can be named.
OCR covers 32 languages and is designed to improve reading under low light, blur or tilt, including specialized vocabulary and long-document structure. Preserve legible source images and ask for fields, quotations or a structured table that can be checked. Text and visual understanding are combined in one model, so a question can refer jointly to a document passage, a figure and the user’s instructions.
Architecture changes
Interleaved-MRoPE distributes positional representation across time, width and height for video reasoning. DeepStack combines visual features from several encoder levels to capture finer details and improve image–text alignment. Text–timestamp alignment grounds events in temporal references. The released text configuration has 94 layers, 128 experts and eight experts selected per token. These architectural details explain the model’s multimodal representation; they do not establish a response-time guarantee.
Preparing visual input and output
The maker’s examples cover single or multiple images, video and text-only requests. Label each visual input so the question can refer to it clearly. Image and video processors use separate pixel budgets: an image budget limits its pixels, while a video budget accounts for pixels across the sampled frames. Explicit image dimensions are aligned to multiples of 32. Video sampling can be controlled by rate or frame count; the documented default is two frames per second. Preserve timestamps when asking for an event timeline.
For a long clip, ask for the event, its time interval and the visible evidence rather than an undifferentiated summary. For several pages, request a result organized by page identifier. These preparation examples apply the documented capabilities and help make omissions or extraction errors visible.
Generation guidance and limits
The checkpoint’s generation file defaults to temperature 0.8, top_p 0.95 and top_k 20. Qwen’s separate Thinking evaluation recipe uses temperature 0.6, presence penalty zero, repetition penalty 1.0 and a 40,960-token output allowance. These are distinct producer presets, not a single hard output cap. Allow space for reasoning and final text alongside the visual tokens, and verify extracted text, calculated answers and generated code against the source before using them.
POST /v1/chat/completionstexttext| Field | Type | Requirement | Values & limits | Default |
|---|---|---|---|---|
Fieldmodel | Typestring | RequirementRequired | Values & limits qwen3-vl-235b-a22b-thinking | Default— |
Fieldmessages | Typemessage[] | RequirementRequired | Values & limits Minimum items: 1 | Default— |
Fieldmessages[].role | Typestring | RequirementRequired | Values & limits Examples: system, user, assistant, tool, function Conditions: These roles are explicitly represented by Source compatibility types; the exact model role set remains model-specific. Model-specific | Default— |
Fieldmessages[].content | Typestring|content_part[]|null | RequirementOptional | Values & limits | Default— |
Fieldmessages[].tool_calls | Typetool_call[] | RequirementOptional | Values & limits | Default— |
Fieldmessages[].tool_call_id | Typestring | RequirementOptional | Values & limits Applies when: role=tool; use actual returned ID | Default— |
Fieldmessages[].name | Typestring | RequirementOptional | Values & limits | Default— |
Fieldmessages[].reasoning_content | Typestring | RequirementOptional | Values & limits Applies when: Round-trip actual returned reasoning where this model uses it | Default— |
Fieldmessages[].reasoning_details | Typeobject|array | RequirementOptional | Values & limits Model-specific | Default— |
Fieldmax_tokens | Typeinteger | RequirementOptional | Values & limits Minimum: 1 Model-specific | Default— |
Fieldmax_completion_tokens | Typeinteger | RequirementOptional | Values & limits Minimum: 1 Model-specific | Default— |
Fieldstream | Typeboolean | RequirementOptional | Values & limits | Defaultfalse |
Fieldstream_options.include_usage | Typeboolean | RequirementOptional | Values & limits Model-specific | Default— |
Fieldtemperature | Typenumber | RequirementOptional | Values & limits Model-specific | Default— |
Fieldtop_p | Typenumber | RequirementOptional | Values & limits Model-specific | Default— |
Fieldstop | Typestring|string[] | RequirementOptional | Values & limits Model-specific | Default— |
Fieldtools | Typefunction_tool[] | RequirementOptional | Values & limits Model-specific | Default— |
Fieldtools[].type | Typestring | RequirementOptional | Values & limits function Model-specific | Default— |
Fieldtools[].function.name | Typestring | RequirementRequired | Values & limits Applies when: function tool supplied | Default— |
Fieldtools[].function.description | Typestring | RequirementOptional | Values & limits | Default— |
Fieldtools[].function.parameters | TypeJSON_schema | RequirementOptional | Values & limits | Default— |
Fieldtool_choice | Typestring|object | RequirementOptional | Values & limits Model-specific | Default— |
Fieldparallel_tool_calls | Typeboolean | RequirementOptional | Values & limits Model-specific | Default— |
Fieldresponse_format | Typeobject | RequirementOptional | Values & limits Model-specific | Default— |
Fieldreasoning_effort | Typestring | RequirementOptional | Values & limits Model-specific | Default— |
Fieldreasoning | Typeobject | RequirementOptional | Values & limits Model-specific | Default— |
Fieldservice_tier | Typestring | RequirementOptional | Values & limits Model-specific | Default— |
Fieldinstructions | Typestring | RequirementOptional | Values & limits Compatibility field | Default— |
Fieldfunctions | Typefunction[] | RequirementOptional | Values & limits Model-specific | Default— |
Fieldfunction_call | Typestring|object | RequirementOptional | Values & limits Model-specific | Default— |
Fieldprompt_cache_options | Typeobject | RequirementOptional | Values & limits Model-specific | Default— |
Fieldmodalities | Typestring[] | RequirementOptional | Values & limits text | Default— |
Fieldprovider | Typeobject | RequirementOptional | Values & limits Conditions: provider.options is rejected on this Ultra text route | Default— |
Fieldcache_control.ttl | Typestring | RequirementOptional | Values & limits Conditions: 1h is rejected in actual prompt-cache locations | Default— |
Text · UTF8 text
messages[].content
| Parameter | Contract |
|---|---|
Parametercontent_type | Contractapplication/json |
Parameterrequired | Contractmodel, messages |
Parametermaterial_types | Contracttext |
Parametermessages | Contractmessage array with role/content shape |
Parameterfile_material | Contractactual UTF-8 document inserted as literal text; no file ID inferred |
Parameteroptional_controls | Contractstream, temperature, top_p, stop, max_tokens, max_completion_tokens, reasoning, response_format, tools, tool_choice |
Parametercontrol_scope | Contractmodel-specific; user function tools differ from hosted tools |
Parameteroutput_modalities | Contracttext |
Parameterhosted_operations | Contractfalse |
Parametermultimodal_input_binding | Contractnull |
Lifecycle: synchronous_or_streamed
Send exact model ID with messages.
Context length is not an output cap.
A text file may supply literal text; image/audio/file-object inputs need exact binding.
Tariff references do not add request options. Supported parameters are listed for the selected method.
| Configuration | Rate schedule | Charge |
|---|---|---|
| ConfigurationBase token rates | Rate scheduleBase rate | Charge Input · Input tokens $0.000572 / 1K tokens Output · Output tokens $0.00572 / 1K tokens Input · Cache read $0 / 1K tokens Input · Cache write $0 / 1K tokens Input · Cache write, 1 hour — / 1K tokens |
Ailoot reserves an estimated amount before a request. After the result, it settles the final charge from measured usage. Streaming settlement uses the final usage record. How reservation and settlement work →
set -eu\n: "${AILOOT_API_KEY:?Set your Ailoot API key}"\npython3 - <<'PY_REQUEST'\nimport json,os\nfrom pathlib import Path\np=json.loads('{"model": "qwen3-vl-235b-a22b-thinking", "messages": [{"role": "user", "content": "Explain the difference between an observation and a hypothesis in three sentences."}]}')\n\nPath("request.json").write_text(json.dumps(p,ensure_ascii=False))\nPY_REQUEST\ncurl --fail-with-body --silent --show-error https://api.ailoot.ai/v1/chat/completions \\\n -H "Authorization: Bearer $AILOOT_API_KEY" -H 'Content-Type: application/json' \\\n --data-binary @request.json -D response-headers.txt -o result.json\npython3 - <<'PY_RESULT'\nimport json\nfrom pathlib import Path\nr=json.loads(Path("result.json").read_text())\nfor x in r.get("choices",[]):\n if x.get("message",{}).get("content"): print(x["message"]["content"])\nPY_RESULT\nfrom openai import OpenAI\nimport os\n\nif not os.environ.get("AILOOT_API_KEY", "").strip(): raise SystemExit("Set AILOOT_API_KEY before making a request")\nclient=OpenAI(base_url="https://api.ailoot.ai/v1",api_key=os.environ["AILOOT_API_KEY"],max_retries=0)\nresult=client.chat.completions.create(model='qwen3-vl-235b-a22b-thinking',messages=[{"role":"user","content":"Explain observations and hypotheses in three sentences."}])\nimport base64, json\nfrom pathlib import Path\n\ndef save_result(data, stem="result"):\n # Keep every field, including usage, IDs, reasoning and tool requests.\n Path(stem + ".json").write_text(json.dumps(data, ensure_ascii=False, indent=2), encoding="utf-8")\n print(json.dumps(data, ensure_ascii=False, indent=2))\n def save_blob(encoded, suffix, label):\n raw = base64.b64decode(encoded, altchars=b"-_", validate=True)\n ext = "png" if raw.startswith(b"\\x89PNG\\r\\n\\x1a\\n") else "jpg" if raw.startswith(b"\\xff\\xd8") else "webp" if raw[:4] == b"RIFF" and raw[8:12] == b"WEBP" else suffix\n file = Path(stem + "-" + label + "." + ext)\n file.write_bytes(raw)\n print("Saved", file)\n for index, block in enumerate(data.get("content") or []):\n print("Content", index, json.dumps(block, ensure_ascii=False))\n for choice in data.get("choices") or []:\n print("Choice", choice.get("index"), json.dumps(choice.get("message", choice.get("delta", {})), ensure_ascii=False))\n for index, item in enumerate(data.get("output") or []):\n print("Output", index, json.dumps(item, ensure_ascii=False))\n if item.get("type") == "image_generation_call" and item.get("result"):\n save_blob(item["result"], item.get("output_format") or "bin", "output-" + str(index))\n for ci, candidate in enumerate(data.get("candidates") or []):\n for pi, part in enumerate((candidate.get("content") or {}).get("parts") or []):\n print("Candidate/part", ci, pi, json.dumps(part, ensure_ascii=False))\n blob = part.get("inlineData") or part.get("inline_data")\n if blob and blob.get("data"):\n mime = blob.get("mimeType") or blob.get("mime_type")\n suffix = {"image/png":"png", "image/jpeg":"jpg", "image/webp":"webp", "audio/wav":"wav", "audio/mpeg":"mp3", "video/mp4":"mp4"}.get(mime, "bin")\n save_blob(blob["data"], suffix, "candidate-" + str(ci) + "-part-" + str(pi))\n\nsave_result(result.to_dict(mode="json"))\nfrom openai import OpenAI\nimport os\nfrom pathlib import Path\n\nif not os.environ.get("AILOOT_API_KEY", "").strip(): raise SystemExit("Set AILOOT_API_KEY before making a request")\nclient=OpenAI(base_url="https://api.ailoot.ai/v1",api_key=os.environ["AILOOT_API_KEY"],max_retries=0)\nimport base64, json\nfrom pathlib import Path\n\ndef save_result(data, stem="result"):\n # Keep every field, including usage, IDs, reasoning and tool requests.\n Path(stem + ".json").write_text(json.dumps(data, ensure_ascii=False, indent=2), encoding="utf-8")\n print(json.dumps(data, ensure_ascii=False, indent=2))\n def save_blob(encoded, suffix, label):\n raw = base64.b64decode(encoded, altchars=b"-_", validate=True)\n ext = "png" if raw.startswith(b"\\x89PNG\\r\\n\\x1a\\n") else "jpg" if raw.startswith(b"\\xff\\xd8") else "webp" if raw[:4] == b"RIFF" and raw[8:12] == b"WEBP" else suffix\n file = Path(stem + "-" + label + "." + ext)\n file.write_bytes(raw)\n print("Saved", file)\n for index, block in enumerate(data.get("content") or []):\n print("Content", index, json.dumps(block, ensure_ascii=False))\n for choice in data.get("choices") or []:\n print("Choice", choice.get("index"), json.dumps(choice.get("message", choice.get("delta", {})), ensure_ascii=False))\n for index, item in enumerate(data.get("output") or []):\n print("Output", index, json.dumps(item, ensure_ascii=False))\n if item.get("type") == "image_generation_call" and item.get("result"):\n save_blob(item["result"], item.get("output_format") or "bin", "output-" + str(index))\n for ci, candidate in enumerate(data.get("candidates") or []):\n for pi, part in enumerate((candidate.get("content") or {}).get("parts") or []):\n print("Candidate/part", ci, pi, json.dumps(part, ensure_ascii=False))\n blob = part.get("inlineData") or part.get("inline_data")\n if blob and blob.get("data"):\n mime = blob.get("mimeType") or blob.get("mime_type")\n suffix = {"image/png":"png", "image/jpeg":"jpg", "image/webp":"webp", "audio/wav":"wav", "audio/mpeg":"mp3", "video/mp4":"mp4"}.get(mime, "bin")\n save_blob(blob["data"], suffix, "candidate-" + str(ci) + "-part-" + str(pi))\n\nwith client.chat.completions.stream(model='qwen3-vl-235b-a22b-thinking',messages=[{"role":"user","content":"Explain observations and hypotheses in three sentences."}]) as stream:\n with Path("stream-events.jsonl").open("w",encoding="utf-8") as log:\n for event in stream:\n payload=event.to_dict(mode="json")\n log.write(json.dumps(payload,ensure_ascii=False)+"\\n"); log.flush()\n print(json.dumps(payload,ensure_ascii=False))\n final=stream.get_final_completion()\n save_result(final.to_dict(mode="json"),"stream-result")\njson or SSE
Keep every complete returned response, file and content part.
Keep every choice and its content/tool calls/reasoning. In SSE assemble deltas by choice/tool index, retain the final finish reason and usage.
Keep request/response IDs and headers. No generic text job retrieval URL is supplied.
| Field | Type |
|---|---|
Fieldchoices[].index | Typeinteger |
Fieldchoices[].message.content | Typestring|null |
Fieldchoices[].message.tool_calls | Typetool_call[] |
Fieldchoices[].message.reasoning_content | Typestring |
Fieldchoices[].message.reasoning_details | Typeobject|array |
Fieldchoices[].finish_reason | Typestring|null |
Fieldusage | Typeobject |
| Response | Condition | Next step |
|---|---|---|
Response400 · invalid_request_error | ConditionInvalid required fields, material representation or parameter value | Next stepCorrect the indicated fields before a new creation request. |
Ailoot Ultra qwen/qwen3-vl-235b-a22b-thinking, qwen3-vl-235b-a22b-thinking
/chat/completions
qwen/qwen3-vl-235b-a22b-thinking, qwen3-vl-235b-a22b-thinking
qwen3-vl-235b-a22b-thinkingQwen3-VL-235B-A22B-Thinking is Qwen’s mixture-of-experts vision-language model for text, image and video reasoning. Use the Thinking edition for STEM questions, document evidence, spatial relationships, visual coding and timestamped video analysis, with a native 262,144-token context.
This description covers the model’s capabilities. Available in Ailoot inputs, outputs and options are listed for the selected method below.
Introduction: the multimodal Thinking edition
Qwen3-VL-235B-A22B-Thinking is the reasoning-focused mixture-of-experts edition of Qwen3-VL. It combines text with images and video, producing textual answers or code. The fixed Thinking checkpoint is separate from Qwen3-VL-235B-A22B-Instruct and the separately named FP8 builds. Choose it for a question that requires connecting visual evidence with a longer explanation, rather than assuming every Qwen3-VL edition has the same response style.
Visual agents and visual coding
Qwen describes a visual-agent workflow that recognizes elements in computer or mobile interfaces, interprets their functions and works with tools to complete tasks. The model supplies perception and reasoning; an application must connect and execute the actions. For an interface review, provide the current screenshot and a specific goal, then ask it to identify the relevant controls and explain the next action. The image alone does not confer control of the device.
Visual coding includes generating diagram content and HTML, CSS or JavaScript from images and videos. Supply a screenshot with the required layout, behavior and editable output format. Ask for the page structure and styles as separate artifacts when that makes review easier. Generated code is text for a development workflow; it is not evidence that a website has been rendered or that the source image has become a working application.
Spatial perception
The maker documents reasoning about object positions, viewpoints and occlusions, with two-dimensional grounding and support for three-dimensional grounding. Ask a concrete relationship question and identify the object of interest: for example, which component is obscured, what is in front of it, and which visual clues support that interpretation. Grounding capability should not be treated as a calibrated physical measurement from an arbitrary picture.
Long context and video
The native context is 262,144 tokens. Qwen describes expansion to roughly one million through YaRN configuration, and family workflows involving books or hours-long video with event indexing at second-level granularity. These capabilities depend on representation, frame sampling and context configuration; the broad maker description does not guarantee perfect recall of every frame. In its repository, Qwen advises an extrapolation factor of two or three for this interleaved positional scheme when extending 256K to about 1M, rather than automatically copying a factor of four from another model.
Multimodal reasoning and recognition
The Thinking edition targets evidence-based STEM and mathematical reasoning, causal analysis and logical answers that connect text with visual material. Give a diagram and its definitions together, or a chart and the precise numerical question. Request the labels and values used before the conclusion. Qwen also describes broader visual recognition across people, anime, products, landmarks, plants and animals. A recognition answer remains an inference from the supplied image, not a guarantee that every object or identity can be named.
OCR covers 32 languages and is designed to improve reading under low light, blur or tilt, including specialized vocabulary and long-document structure. Preserve legible source images and ask for fields, quotations or a structured table that can be checked. Text and visual understanding are combined in one model, so a question can refer jointly to a document passage, a figure and the user’s instructions.
Architecture changes
Interleaved-MRoPE distributes positional representation across time, width and height for video reasoning. DeepStack combines visual features from several encoder levels to capture finer details and improve image–text alignment. Text–timestamp alignment grounds events in temporal references. The released text configuration has 94 layers, 128 experts and eight experts selected per token. These architectural details explain the model’s multimodal representation; they do not establish a response-time guarantee.
Preparing visual input and output
The maker’s examples cover single or multiple images, video and text-only requests. Label each visual input so the question can refer to it clearly. Image and video processors use separate pixel budgets: an image budget limits its pixels, while a video budget accounts for pixels across the sampled frames. Explicit image dimensions are aligned to multiples of 32. Video sampling can be controlled by rate or frame count; the documented default is two frames per second. Preserve timestamps when asking for an event timeline.
For a long clip, ask for the event, its time interval and the visible evidence rather than an undifferentiated summary. For several pages, request a result organized by page identifier. These preparation examples apply the documented capabilities and help make omissions or extraction errors visible.
Generation guidance and limits
The checkpoint’s generation file defaults to temperature 0.8, top_p 0.95 and top_k 20. Qwen’s separate Thinking evaluation recipe uses temperature 0.6, presence penalty zero, repetition penalty 1.0 and a 40,960-token output allowance. These are distinct producer presets, not a single hard output cap. Allow space for reasoning and final text alongside the visual tokens, and verify extracted text, calculated answers and generated code against the source before using them.
POST /v1/messagestextjson, sse| Field | Type | Requirement | Values & limits | Default |
|---|---|---|---|---|
Fieldmodel | Typestring | RequirementRequired | Values & limits qwen3-vl-235b-a22b-thinking | Default— |
Fieldmessages | Typemessage[] | RequirementRequired | Values & limits | Default— |
Fieldmessages[].role | Typestring | RequirementRequired | Values & limits user, assistant | Default— |
Fieldmessages[].content | Typestring | RequirementRequired | Values & limits | Default— |
Fieldmax_tokens | Typeinteger | RequirementRequired | Values & limits | Default— |
Fieldstream | Typeboolean | RequirementOptional | Values & limits | Defaultfalse |
Text · UTF-8 text
messages[].content string
| Parameter | Contract |
|---|---|
Parametercontent_type | Contractapplication/json |
Parameterrequired | Contractmodel, messages, max_tokens |
Parameteroptional_controls | Contractstream |
Parametermaterial_types | Contracttext |
Parametertext_file_material | ContractRead the actual UTF-8 document and put its text in the indicated input or message content string. |
Lifecycle: synchronous_or_streamed
Tariff references do not add request options. Supported parameters are listed for the selected method.
| Configuration | Rate schedule | Charge |
|---|---|---|
| ConfigurationBase token rates | Rate scheduleBase rate | Charge Input · Input tokens $0.000572 / 1K tokens Output · Output tokens $0.00572 / 1K tokens Input · Cache read $0 / 1K tokens Input · Cache write $0 / 1K tokens Input · Cache write, 1 hour — / 1K tokens |
Ailoot reserves an estimated amount before a request. After the result, it settles the final charge from measured usage. Streaming settlement uses the final usage record. How reservation and settlement work →
{\n "model": "qwen3-vl-235b-a22b-thinking",\n "max_tokens": 128,\n "messages": [\n {\n "role": "user",\n "content": "Summarize the incident report."\n }\n ],\n "stream": false\n}set -eu\n: "${AILOOT_API_KEY:?Set your Ailoot API key}"\n: "${AILOOT_TEXT_FILE:?Set your actual UTF-8 document}"\nexport AILOOT_TEXT_FILE\npython3 - <<'PY_REQUEST'\nimport json, os\nfrom pathlib import Path\n\nsource_file = Path(os.environ['AILOOT_TEXT_FILE'])\nif not source_file.is_file():\n raise SystemExit('Set AILOOT_TEXT_FILE to your actual UTF-8 document')\ntext = source_file.read_text(encoding='utf-8')\nif not text.strip():\n raise SystemExit('Choose a document containing text')\npayload = {'model': 'qwen3-vl-235b-a22b-thinking', 'max_tokens': 128, 'messages': [{'role': 'user', 'content': 'Summarize the incident report.'}]}\npayload['messages'][0]['content'] = 'Summarize this material and quote supporting source passages.\\n\\n' + text\nPath('request.json').write_text(json.dumps(payload, ensure_ascii=False), encoding='utf-8')\nPY_REQUEST\ncurl --fail-with-body --silent --show-error https://api.ailoot.ai/v1/messages \\\n -H "x-api-key: $AILOOT_API_KEY" -H 'anthropic-version: 2023-06-01' -H 'Content-Type: application/json' \\\n --data-binary @request.json -D response-headers.txt -o result.json\npython3 - <<'PY_RESULT'\nimport json\nfrom pathlib import Path\n\nresult = json.loads(Path('result.json').read_text(encoding='utf-8'))\nprint(json.dumps(result, ensure_ascii=False, indent=2))\nPY_RESULT\nimport json, os\nfrom pathlib import Path\n\nsource_file = Path(os.environ['AILOOT_TEXT_FILE'])\nif not source_file.is_file():\n raise SystemExit('Set AILOOT_TEXT_FILE to your actual UTF-8 document')\ntext = source_file.read_text(encoding='utf-8')\nif not text.strip():\n raise SystemExit('Choose a document containing text')\npayload = {'model': 'qwen3-vl-235b-a22b-thinking', 'max_tokens': 128, 'messages': [{'role': 'user', 'content': 'Summarize the incident report.'}]}\npayload['messages'][0]['content'] = 'Summarize this material and quote supporting source passages.\\n\\n' + text\nfrom urllib.request import Request, urlopen\nfrom urllib.error import HTTPError\n\napi_key = os.environ['AILOOT_API_KEY']\nif not api_key.strip():\n raise SystemExit('Set AILOOT_API_KEY before making a request')\nheaders = {'x-api-key': 'PLACEHOLDER_KEY', 'anthropic-version': '2023-06-01', 'Content-Type': 'application/json'}\nheaders['x-api-key'] = api_key\nrequest = Request('https://api.ailoot.ai/v1/messages', data=json.dumps(payload, ensure_ascii=False).encode('utf-8'), headers=headers, method='POST')\ntry:\n with urlopen(request, timeout=120) as response:\n raw = response.read()\n Path('response-headers.txt').write_text(str(response.headers), encoding='utf-8')\nexcept HTTPError as error:\n Path('result.json').write_bytes(error.read())\n Path('response-headers.txt').write_text(str(error.headers), encoding='utf-8')\n raise\nPath('result.json').write_bytes(raw)\nimport json\nfrom pathlib import Path\n\nresult = json.loads(Path('result.json').read_text(encoding='utf-8'))\nprint(json.dumps(result, ensure_ascii=False, indent=2))\nJSON or SSE
Keep every complete returned response, file and content part.
Keep all content blocks, thinking signatures and actual tool-use IDs. Assemble all SSE block deltas; a first text block is not the complete output.
Keep request/response IDs and headers. No generic text job retrieval URL is supplied.
| Field | Type |
|---|---|
Fieldid | Typestring |
Fieldmodel | Typestring |
Fieldcontent[] | Typetext|thinking|tool_use|other content block |
Fieldstop_reason | Typestring|null |
Fieldusage | Typeobject |
| Response | Condition | Next step |
|---|---|---|
Response400 · invalid_request_error | ConditionInvalid required fields, material representation or parameter value | Next stepCorrect the indicated fields before a new creation request. |
Ailoot Ultra qwen/qwen3-vl-235b-a22b-thinking, qwen3-vl-235b-a22b-thinking
/chat/completions
qwen/qwen3-vl-235b-a22b-thinking, qwen3-vl-235b-a22b-thinking
Technical example in English
Reason from a textual circuit description\n\nUse this textual circuit description: an ideal 12 V source powers two series resistors, R1 = 2 Ω and R2 = 4 Ω. Calculate total resistance, current and voltage across each resistor. State the ideal-component assumptions and verify that the voltage drops sum to the source voltage. Return a compact worked solution using the labels R1 and R2.A labeled worked calculation and a voltage-sum consistency check.The same series current is used for both resistors and the stated voltage drops sum to 12 V.