Qwen2-VL-72B-Instruct is the instruction-tuned 72B visual-language edition built around Qwen2. It receives text together with images or video and generates a text answer. The maker highlights understanding images with different resolutions and aspect ratios, interpreting text inside images, reasoning about visible relationships and answering questions about videos longer than 20 minutes.
Image and video capabilities
For a document page, ask for the particular content you need: a paragraph, a table row, a formula or the conditions attached to a statement. Keep the page readable and name whether the result should transcribe the wording, summarize it or explain its meaning. A useful document output can retain the original section label beside the extracted content. For a diagram, ask how the labeled elements connect and identify the relationship that needs explanation.
The documented video use includes questions, dialogue and written content based on the visible sequence. Define the event or change to inspect: when an object appears, what happens before a particular step, or how a demonstrated action relates to a written procedure. Request an ordered event description or a comparison between the visible steps and the supplied procedure. The maker's 20-minute-plus statement describes long-video understanding; it is not a fixed maximum video duration or a guarantee that every raw clip fits an unchanged input configuration.
Visual-agent use
The release also positions Qwen2-VL for visual agents connected to mobile devices, robots and other tools. Its announcement describes function calling and code-interpreter scenarios, such as producing code from a diagram or using a visual question to request an external lookup. The surrounding application supplies those tools, executes the proposed action and returns the result. The model's text response alone does not click a control, move a device or execute the generated program.
A practical interface task can supply a screenshot and the user's goal, then ask for the visible control relevant to the next step and the evidence supporting that choice. Keep observation, proposed action and actual execution as separate parts of the workflow. This uses the visual-agent focus while making the requested result specific enough for the application to act on.
Multilingual text inside images
The maker describes recognition of text inside images across English and Chinese, most European languages, and languages including Japanese, Korean, Arabic and Vietnamese. A multilingual page can therefore be approached as a visual reading task. Request the original transcription first when exact wording matters, then a translation or explanation in the chosen language. Preserve labels and numbers rather than mixing recognition with an unspecified rewrite.
Dynamic resolution and positional information
Qwen2-VL combines a vision encoder with the Qwen2 language model. Naive Dynamic Resolution maps different image sizes into different numbers of visual tokens. Multimodal Rotary Position Embedding separates positional information so the model can represent text order, two-dimensional images and the temporal and spatial structure of video. These features explain how the same family handles a page, several images or a video sequence.
The model card's image processor documents a default range of 4 to 16,384 visual tokens per image. The application can adjust minimum and maximum pixels while preserving aspect ratio, or request dimensions that are rounded to a multiple of 28. The example range of 256 to 1,280 tokens is a configurable illustration, not a separate universal image limit. More image detail uses more computation and visual tokens, so retain the labels, headings and small objects needed by the question when preparing the image.
Input preparation and context
The maker demonstrates a single image, multiple images, video supplied as a frame sequence or local file, and batched conversations. Image preparation includes local files, encoded image data and image references. The card's video example uses a selected frame rate; that example does not establish one required sampling rate for every video. Identify the order and role of multiple images, and provide the information needed to interpret a sequence rather than treating unrelated frames as one scene.
The released configuration sets maximum positions to 32,768. Visual material and textual instructions must be planned within the actual configured context. The card's 128-token generation example is only an example setting; it does not publish a separate maximum output capacity. The 72B edition should retain its own configuration rather than inheriting the long-context settings of Qwen2.5-VL or a different Qwen model.
Substantive limitations
The maker explicitly states that Qwen2-VL does not understand audio contained in videos. Spoken information must therefore be provided separately if the question depends on it. The image-training dataset is described as updated through June 2023; this is a statement about that dataset rather than a promise of current visual knowledge.
The documented limitations also include incomplete recognition of particular people or named intellectual properties, difficulty with intricate multi-step instructions, weak counting in complex scenes and limitations in three-dimensional spatial reasoning. When the task depends on one of these abilities, focus the question on the required objects or relationship and preserve the visual evidence beside the answer. Clear source material and a defined output make the result easier to inspect without replacing those stated limitations with an accuracy guarantee.
", "aboutText": "Qwen2-VL-72B-Instruct is the instruction-tuned 72B visual-language edition built around Qwen2. It receives text together with images or video and generates a text answer. The maker highlights understanding images with different resolutions and aspect ratios, interpreting text inside images, reasoning about visible relationships and answering questions about videos longer than 20 minutes.\n\nImage and video capabilities\n\nFor a document page, ask for the particular content you need: a paragraph, a table row, a formula or the conditions attached to a statement. Keep the page readable and name whether the result should transcribe the wording, summarize it or explain its meaning. A useful document output can retain the original section label beside the extracted content. For a diagram, ask how the labeled elements connect and identify the relationship that needs explanation.\n\nThe documented video use includes questions, dialogue and written content based on the visible sequence. Define the event or change to inspect: when an object appears, what happens before a particular step, or how a demonstrated action relates to a written procedure. Request an ordered event description or a comparison between the visible steps and the supplied procedure. The maker's 20-minute-plus statement describes long-video understanding; it is not a fixed maximum video duration or a guarantee that every raw clip fits an unchanged input configuration.\n\nVisual-agent use\n\nThe release also positions Qwen2-VL for visual agents connected to mobile devices, robots and other tools. Its announcement describes function calling and code-interpreter scenarios, such as producing code from a diagram or using a visual question to request an external lookup. The surrounding application supplies those tools, executes the proposed action and returns the result. The model's text response alone does not click a control, move a device or execute the generated program.\n\nA practical interface task can supply a screenshot and the user's goal, then ask for the visible control relevant to the next step and the evidence supporting that choice. Keep observation, proposed action and actual execution as separate parts of the workflow. This uses the visual-agent focus while making the requested result specific enough for the application to act on.\n\nMultilingual text inside images\n\nThe maker describes recognition of text inside images across English and Chinese, most European languages, and languages including Japanese, Korean, Arabic and Vietnamese. A multilingual page can therefore be approached as a visual reading task. Request the original transcription first when exact wording matters, then a translation or explanation in the chosen language. Preserve labels and numbers rather than mixing recognition with an unspecified rewrite.\n\nDynamic resolution and positional information\n\nQwen2-VL combines a vision encoder with the Qwen2 language model. Naive Dynamic Resolution maps different image sizes into different numbers of visual tokens. Multimodal Rotary Position Embedding separates positional information so the model can represent text order, two-dimensional images and the temporal and spatial structure of video. These features explain how the same family handles a page, several images or a video sequence.\n\nThe model card's image processor documents a default range of 4 to 16,384 visual tokens per image. The application can adjust minimum and maximum pixels while preserving aspect ratio, or request dimensions that are rounded to a multiple of 28. The example range of 256 to 1,280 tokens is a configurable illustration, not a separate universal image limit. More image detail uses more computation and visual tokens, so retain the labels, headings and small objects needed by the question when preparing the image.\n\nInput preparation and context\n\nThe maker demonstrates a single image, multiple images, video supplied as a frame sequence or local file, and batched conversations. Image preparation includes local files, encoded image data and image references. The card's video example uses a selected frame rate; that example does not establish one required sampling rate for every video. Identify the order and role of multiple images, and provide the information needed to interpret a sequence rather than treating unrelated frames as one scene.\n\nThe released configuration sets maximum positions to 32,768. Visual material and textual instructions must be planned within the actual configured context. The card's 128-token generation example is only an example setting; it does not publish a separate maximum output capacity. The 72B edition should retain its own configuration rather than inheriting the long-context settings of Qwen2.5-VL or a different Qwen model.\n\nSubstantive limitations\n\nThe maker explicitly states that Qwen2-VL does not understand audio contained in videos. Spoken information must therefore be provided separately if the question depends on it. The image-training dataset is described as updated through June 2023; this is a statement about that dataset rather than a promise of current visual knowledge.\n\nThe documented limitations also include incomplete recognition of particular people or named intellectual properties, difficulty with intricate multi-step instructions, weak counting in complex scenes and limitations in three-dimensional spatial reasoning. When the task depends on one of these abilities, focus the question on the required objects or relationship and preserve the visual evidence beside the answer. Clear source material and a defined output make the result easier to inspect without replacing those stated limitations with an accuracy guarantee.", "editorialOverlay": { "copy_fields": { "practical_html": "Technical example in English
Read a supplied document transcription\n\nUse this text transcription of page P2, section Limits: 'Store below 25 C. Once opened, use within seven days. Do not freeze.' Return the original wording first, then a plain-English explanation separating the temperature, time and freezing conditions. Preserve P2 as the source label. No image or video is attached.A labelled transcription followed by three source-grounded explanations of the conditions.Below 25 C is not rewritten as at or below 25 C, and the seven-day period starts at opening.Qwen/Qwen2-VL-72B-InstructQwen2-VL-72B-Instruct combines text, images and video for visual questions, document reading and multilingual text in images. Its maker describes video understanding beyond 20 minutes, dynamic image resolution and visual-agent integrations; the answer is text.
This description covers the model’s capabilities. Available in Ailoot inputs, outputs and options are listed for the selected method below.
Qwen2-VL-72B-Instruct is the instruction-tuned 72B visual-language edition built around Qwen2. It receives text together with images or video and generates a text answer. The maker highlights understanding images with different resolutions and aspect ratios, interpreting text inside images, reasoning about visible relationships and answering questions about videos longer than 20 minutes.
Image and video capabilities
For a document page, ask for the particular content you need: a paragraph, a table row, a formula or the conditions attached to a statement. Keep the page readable and name whether the result should transcribe the wording, summarize it or explain its meaning. A useful document output can retain the original section label beside the extracted content. For a diagram, ask how the labeled elements connect and identify the relationship that needs explanation.
The documented video use includes questions, dialogue and written content based on the visible sequence. Define the event or change to inspect: when an object appears, what happens before a particular step, or how a demonstrated action relates to a written procedure. Request an ordered event description or a comparison between the visible steps and the supplied procedure. The maker's 20-minute-plus statement describes long-video understanding; it is not a fixed maximum video duration or a guarantee that every raw clip fits an unchanged input configuration.
Visual-agent use
The release also positions Qwen2-VL for visual agents connected to mobile devices, robots and other tools. Its announcement describes function calling and code-interpreter scenarios, such as producing code from a diagram or using a visual question to request an external lookup. The surrounding application supplies those tools, executes the proposed action and returns the result. The model's text response alone does not click a control, move a device or execute the generated program.
A practical interface task can supply a screenshot and the user's goal, then ask for the visible control relevant to the next step and the evidence supporting that choice. Keep observation, proposed action and actual execution as separate parts of the workflow. This uses the visual-agent focus while making the requested result specific enough for the application to act on.
Multilingual text inside images
The maker describes recognition of text inside images across English and Chinese, most European languages, and languages including Japanese, Korean, Arabic and Vietnamese. A multilingual page can therefore be approached as a visual reading task. Request the original transcription first when exact wording matters, then a translation or explanation in the chosen language. Preserve labels and numbers rather than mixing recognition with an unspecified rewrite.
Dynamic resolution and positional information
Qwen2-VL combines a vision encoder with the Qwen2 language model. Naive Dynamic Resolution maps different image sizes into different numbers of visual tokens. Multimodal Rotary Position Embedding separates positional information so the model can represent text order, two-dimensional images and the temporal and spatial structure of video. These features explain how the same family handles a page, several images or a video sequence.
The model card's image processor documents a default range of 4 to 16,384 visual tokens per image. The application can adjust minimum and maximum pixels while preserving aspect ratio, or request dimensions that are rounded to a multiple of 28. The example range of 256 to 1,280 tokens is a configurable illustration, not a separate universal image limit. More image detail uses more computation and visual tokens, so retain the labels, headings and small objects needed by the question when preparing the image.
Input preparation and context
The maker demonstrates a single image, multiple images, video supplied as a frame sequence or local file, and batched conversations. Image preparation includes local files, encoded image data and image references. The card's video example uses a selected frame rate; that example does not establish one required sampling rate for every video. Identify the order and role of multiple images, and provide the information needed to interpret a sequence rather than treating unrelated frames as one scene.
The released configuration sets maximum positions to 32,768. Visual material and textual instructions must be planned within the actual configured context. The card's 128-token generation example is only an example setting; it does not publish a separate maximum output capacity. The 72B edition should retain its own configuration rather than inheriting the long-context settings of Qwen2.5-VL or a different Qwen model.
Substantive limitations
The maker explicitly states that Qwen2-VL does not understand audio contained in videos. Spoken information must therefore be provided separately if the question depends on it. The image-training dataset is described as updated through June 2023; this is a statement about that dataset rather than a promise of current visual knowledge.
The documented limitations also include incomplete recognition of particular people or named intellectual properties, difficulty with intricate multi-step instructions, weak counting in complex scenes and limitations in three-dimensional spatial reasoning. When the task depends on one of these abilities, focus the question on the required objects or relationship and preserve the visual evidence beside the answer. Clear source material and a defined output make the result easier to inspect without replacing those stated limitations with an accuracy guarantee.
POST /v1/chat/completionstexttext| Field | Type | Requirement | Values & limits | Default |
|---|---|---|---|---|
Fieldmodel | Typestring | RequirementRequired | Values & limits Qwen/Qwen2-VL-72B-Instruct | 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 $0.00118 / 1K tokens Output · Output tokens $0.00118 / 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/Qwen2-VL-72B-Instruct", "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='Qwen/Qwen2-VL-72B-Instruct',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='Qwen/Qwen2-VL-72B-Instruct',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/Qwen2-VL-72B-Instruct
Use the method paths shown above.
Qwen/Qwen2-VL-72B-InstructQwen2-VL-72B-Instruct combines text, images and video for visual questions, document reading and multilingual text in images. Its maker describes video understanding beyond 20 minutes, dynamic image resolution and visual-agent integrations; the answer is text.
This description covers the model’s capabilities. Available in Ailoot inputs, outputs and options are listed for the selected method below.
Qwen2-VL-72B-Instruct is the instruction-tuned 72B visual-language edition built around Qwen2. It receives text together with images or video and generates a text answer. The maker highlights understanding images with different resolutions and aspect ratios, interpreting text inside images, reasoning about visible relationships and answering questions about videos longer than 20 minutes.
Image and video capabilities
For a document page, ask for the particular content you need: a paragraph, a table row, a formula or the conditions attached to a statement. Keep the page readable and name whether the result should transcribe the wording, summarize it or explain its meaning. A useful document output can retain the original section label beside the extracted content. For a diagram, ask how the labeled elements connect and identify the relationship that needs explanation.
The documented video use includes questions, dialogue and written content based on the visible sequence. Define the event or change to inspect: when an object appears, what happens before a particular step, or how a demonstrated action relates to a written procedure. Request an ordered event description or a comparison between the visible steps and the supplied procedure. The maker's 20-minute-plus statement describes long-video understanding; it is not a fixed maximum video duration or a guarantee that every raw clip fits an unchanged input configuration.
Visual-agent use
The release also positions Qwen2-VL for visual agents connected to mobile devices, robots and other tools. Its announcement describes function calling and code-interpreter scenarios, such as producing code from a diagram or using a visual question to request an external lookup. The surrounding application supplies those tools, executes the proposed action and returns the result. The model's text response alone does not click a control, move a device or execute the generated program.
A practical interface task can supply a screenshot and the user's goal, then ask for the visible control relevant to the next step and the evidence supporting that choice. Keep observation, proposed action and actual execution as separate parts of the workflow. This uses the visual-agent focus while making the requested result specific enough for the application to act on.
Multilingual text inside images
The maker describes recognition of text inside images across English and Chinese, most European languages, and languages including Japanese, Korean, Arabic and Vietnamese. A multilingual page can therefore be approached as a visual reading task. Request the original transcription first when exact wording matters, then a translation or explanation in the chosen language. Preserve labels and numbers rather than mixing recognition with an unspecified rewrite.
Dynamic resolution and positional information
Qwen2-VL combines a vision encoder with the Qwen2 language model. Naive Dynamic Resolution maps different image sizes into different numbers of visual tokens. Multimodal Rotary Position Embedding separates positional information so the model can represent text order, two-dimensional images and the temporal and spatial structure of video. These features explain how the same family handles a page, several images or a video sequence.
The model card's image processor documents a default range of 4 to 16,384 visual tokens per image. The application can adjust minimum and maximum pixels while preserving aspect ratio, or request dimensions that are rounded to a multiple of 28. The example range of 256 to 1,280 tokens is a configurable illustration, not a separate universal image limit. More image detail uses more computation and visual tokens, so retain the labels, headings and small objects needed by the question when preparing the image.
Input preparation and context
The maker demonstrates a single image, multiple images, video supplied as a frame sequence or local file, and batched conversations. Image preparation includes local files, encoded image data and image references. The card's video example uses a selected frame rate; that example does not establish one required sampling rate for every video. Identify the order and role of multiple images, and provide the information needed to interpret a sequence rather than treating unrelated frames as one scene.
The released configuration sets maximum positions to 32,768. Visual material and textual instructions must be planned within the actual configured context. The card's 128-token generation example is only an example setting; it does not publish a separate maximum output capacity. The 72B edition should retain its own configuration rather than inheriting the long-context settings of Qwen2.5-VL or a different Qwen model.
Substantive limitations
The maker explicitly states that Qwen2-VL does not understand audio contained in videos. Spoken information must therefore be provided separately if the question depends on it. The image-training dataset is described as updated through June 2023; this is a statement about that dataset rather than a promise of current visual knowledge.
The documented limitations also include incomplete recognition of particular people or named intellectual properties, difficulty with intricate multi-step instructions, weak counting in complex scenes and limitations in three-dimensional spatial reasoning. When the task depends on one of these abilities, focus the question on the required objects or relationship and preserve the visual evidence beside the answer. Clear source material and a defined output make the result easier to inspect without replacing those stated limitations with an accuracy guarantee.
POST /v1/messagestextjson, sse| Field | Type | Requirement | Values & limits | Default |
|---|---|---|---|---|
Fieldmodel | Typestring | RequirementRequired | Values & limits Qwen/Qwen2-VL-72B-Instruct | 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 $0.00118 / 1K tokens Output · Output tokens $0.00118 / 1K tokens Input · Cache read $0 / 1K tokens Input · Cache write $0 / 1K tokens Input · Cache write, 1 hour — / 1K tokens |
Ailoot reserves an estimated amount before a request. After the result, it settles the final charge from measured usage. Streaming settlement uses the final usage record. How reservation and settlement work →
{\n "model": "Qwen/Qwen2-VL-72B-Instruct",\n "max_tokens": 128,\n "messages": [\n {\n "role": "user",\n "content": "Summarize the incident report."\n }\n ],\n "stream": false\n}set -eu\n: "${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/Qwen2-VL-72B-Instruct', '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/Qwen2-VL-72B-Instruct', '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/Qwen2-VL-72B-Instruct
Use the method paths shown above.
Technical example in English
Read a supplied document transcription\n\nUse this text transcription of page P2, section Limits: 'Store below 25 C. Once opened, use within seven days. Do not freeze.' Return the original wording first, then a plain-English explanation separating the temperature, time and freezing conditions. Preserve P2 as the source label. No image or video is attached.A labelled transcription followed by three source-grounded explanations of the conditions.Below 25 C is not rewritten as at or below 25 C, and the seven-day period starts at opening.