Introduction: reasoning over long documents
QwenLong-L1-32B is Tongyi Lab’s long-context text reasoning model. Its purpose is to answer questions whose evidence is spread through a document or across several documents, rather than simply repeat a nearby passage. It is built from DeepSeek-R1-Distill-Qwen-32B and further trained for document-grounded reasoning. Supply text and a question; the model generates a textual reasoning sequence and answer. Scanned pages, audio and video need an appropriate separate conversion step before this text workflow.
How the long-context training works
The QwenLong-L1 framework first warms up a reasoning policy with supervised fine-tuning, then increases context length through a reinforcement-learning curriculum. Difficulty-aware retrospective sampling adjusts which examples are revisited as training progresses. The maker combines rule-based and model-based binary outcome rewards, using group-relative policy optimization to encourage answers that remain grounded in long evidence. This describes the checkpoint’s training approach; it does not mean an individual answer is automatically verified by an external judge.
The released edition
This is the original QwenLong-L1-32B release. The separately named QwenLong-L1-32B-AWQ is an INT4-quantized edition and should retain its own identity. The published configuration for this model uses a Qwen2 causal-language-model architecture with 64 layers and bfloat16 weights. The model’s document specialization is more relevant to selection than treating the 32B name as proof that every general-purpose task will improve.
Preparing a document question
The maker’s example separates the document from the question with explicit text delimiters and requests a recognizable final-answer phrase after the reasoning. Follow that structure: provide the relevant document body, keep headings and tables readable, and ask one concrete question. For a report calculation, preserve units, reporting periods and table labels, then request the formula, the values used and a final result. For a policy comparison, label the documents A and B and request a table of matching rules, differences and the supporting section for each row. These are authored applications of the documented numerical and cross-document capabilities.
The example parses reasoning and final-answer text separately. It uses a 10,000-token generation budget, temperature 0.7 and top_p 0.95; those are example settings, not a declared maximum output length. The card also shows an optional identity-setting system message. Preserve this model’s own prompt guidance rather than importing a different checkpoint’s recommendations solely because it appears in the model’s ancestry.
Processing long documents
Tongyi Lab validates context lengths up to 131,072 tokens using YaRN and recommends scaling when combined input and output significantly exceed 32,768 tokens. Its examples describe a factor of four, or a smaller factor such as two for a 65,536-token workload. Static scaling can reduce performance on shorter text, so the card advises against enabling it when ordinary requests stay within 32,768 tokens. The published configuration lists 131,072 positions but no enabled rope_scaling field; configuration and the maker’s conditional long-context procedure should therefore be checked together.
A long input still needs room for reasoning and the answer. Start with the passages needed for the question, remove repeated headers and navigation text, and retain page or section identifiers. Ask for evidence references in the result so a reviewer can trace the conclusion back to the supplied material. A large context allowance helps place more evidence in one request; it does not establish perfect recall of every detail.
What the document training covers
DocQA-RL-1.6K contains 1,600 training problems across three reasoning domains: 600 numerical questions from long specialized documents such as financial reports; 600 logical multiple-choice questions about legal, financial, insurance and production documents; and 400 multi-hop examples that connect evidence across sources. This makes the model a relevant candidate for report arithmetic, rule interpretation and questions requiring several supporting passages. The training domains describe experience with those tasks, not professional certification or guaranteed conclusions.
Practical output and limits
Choose an answer that can be checked: a computed value with units, a rule comparison with section references, or a short conclusion with the chain of supporting facts. If the document omits a required date or definition, ask the model to identify the missing evidence instead of inventing it. The release materials focus on document question answering and do not establish a dedicated tool-calling contract or a separate hard output ceiling. Treat answers as grounded drafts and verify calculations, references and interpretations against the supplied documents.
", "aboutText": "Introduction: reasoning over long documents\n\nQwenLong-L1-32B is Tongyi Lab’s long-context text reasoning model. Its purpose is to answer questions whose evidence is spread through a document or across several documents, rather than simply repeat a nearby passage. It is built from DeepSeek-R1-Distill-Qwen-32B and further trained for document-grounded reasoning. Supply text and a question; the model generates a textual reasoning sequence and answer. Scanned pages, audio and video need an appropriate separate conversion step before this text workflow.\n\nHow the long-context training works\n\nThe QwenLong-L1 framework first warms up a reasoning policy with supervised fine-tuning, then increases context length through a reinforcement-learning curriculum. Difficulty-aware retrospective sampling adjusts which examples are revisited as training progresses. The maker combines rule-based and model-based binary outcome rewards, using group-relative policy optimization to encourage answers that remain grounded in long evidence. This describes the checkpoint’s training approach; it does not mean an individual answer is automatically verified by an external judge.\n\nThe released edition\n\nThis is the original QwenLong-L1-32B release. The separately named QwenLong-L1-32B-AWQ is an INT4-quantized edition and should retain its own identity. The published configuration for this model uses a Qwen2 causal-language-model architecture with 64 layers and bfloat16 weights. The model’s document specialization is more relevant to selection than treating the 32B name as proof that every general-purpose task will improve.\n\nPreparing a document question\n\nThe maker’s example separates the document from the question with explicit text delimiters and requests a recognizable final-answer phrase after the reasoning. Follow that structure: provide the relevant document body, keep headings and tables readable, and ask one concrete question. For a report calculation, preserve units, reporting periods and table labels, then request the formula, the values used and a final result. For a policy comparison, label the documents A and B and request a table of matching rules, differences and the supporting section for each row. These are authored applications of the documented numerical and cross-document capabilities.\n\nThe example parses reasoning and final-answer text separately. It uses a 10,000-token generation budget, temperature 0.7 and top_p 0.95; those are example settings, not a declared maximum output length. The card also shows an optional identity-setting system message. Preserve this model’s own prompt guidance rather than importing a different checkpoint’s recommendations solely because it appears in the model’s ancestry.\n\nProcessing long documents\n\nTongyi Lab validates context lengths up to 131,072 tokens using YaRN and recommends scaling when combined input and output significantly exceed 32,768 tokens. Its examples describe a factor of four, or a smaller factor such as two for a 65,536-token workload. Static scaling can reduce performance on shorter text, so the card advises against enabling it when ordinary requests stay within 32,768 tokens. The published configuration lists 131,072 positions but no enabled rope_scaling field; configuration and the maker’s conditional long-context procedure should therefore be checked together.\n\nA long input still needs room for reasoning and the answer. Start with the passages needed for the question, remove repeated headers and navigation text, and retain page or section identifiers. Ask for evidence references in the result so a reviewer can trace the conclusion back to the supplied material. A large context allowance helps place more evidence in one request; it does not establish perfect recall of every detail.\n\nWhat the document training covers\n\nDocQA-RL-1.6K contains 1,600 training problems across three reasoning domains: 600 numerical questions from long specialized documents such as financial reports; 600 logical multiple-choice questions about legal, financial, insurance and production documents; and 400 multi-hop examples that connect evidence across sources. This makes the model a relevant candidate for report arithmetic, rule interpretation and questions requiring several supporting passages. The training domains describe experience with those tasks, not professional certification or guaranteed conclusions.\n\nPractical output and limits\n\nChoose an answer that can be checked: a computed value with units, a rule comparison with section references, or a short conclusion with the chain of supporting facts. If the document omits a required date or definition, ask the model to identify the missing evidence instead of inventing it. The release materials focus on document question answering and do not establish a dedicated tool-calling contract or a separate hard output ceiling. Treat answers as grounded drafts and verify calculations, references and interpretations against the supplied documents.", "editorialOverlay": { "copy_fields": { "practical_html": "Technical example in English
Connect evidence across report sections\n\nDocument section 1 says 240 units were shipped. Section 4 says 30 shipped units were returned before month end. Appendix A says net shipped units equal shipped minus returned units. Calculate net shipped units for the month, cite every section needed and state whether revenue can be calculated from these facts. End with a clearly labelled final answer.A source-linked formula and calculation, a revenue-information limit and a labelled final result.Check that all three source sections support the calculation and that a price or revenue value is not invented.Tongyi-Zhiwen/QwenLong-L1-32BQwenLong-L1-32B is Tongyi Lab’s text reasoning model for questions grounded in long documents. It combines a DeepSeek-R1-Distill-Qwen-32B starting point with document-focused reinforcement learning for numerical, logical and cross-document reasoning.
This description covers the model’s capabilities. Available in Ailoot inputs, outputs and options are listed for the selected method below.
Introduction: reasoning over long documents
QwenLong-L1-32B is Tongyi Lab’s long-context text reasoning model. Its purpose is to answer questions whose evidence is spread through a document or across several documents, rather than simply repeat a nearby passage. It is built from DeepSeek-R1-Distill-Qwen-32B and further trained for document-grounded reasoning. Supply text and a question; the model generates a textual reasoning sequence and answer. Scanned pages, audio and video need an appropriate separate conversion step before this text workflow.
How the long-context training works
The QwenLong-L1 framework first warms up a reasoning policy with supervised fine-tuning, then increases context length through a reinforcement-learning curriculum. Difficulty-aware retrospective sampling adjusts which examples are revisited as training progresses. The maker combines rule-based and model-based binary outcome rewards, using group-relative policy optimization to encourage answers that remain grounded in long evidence. This describes the checkpoint’s training approach; it does not mean an individual answer is automatically verified by an external judge.
The released edition
This is the original QwenLong-L1-32B release. The separately named QwenLong-L1-32B-AWQ is an INT4-quantized edition and should retain its own identity. The published configuration for this model uses a Qwen2 causal-language-model architecture with 64 layers and bfloat16 weights. The model’s document specialization is more relevant to selection than treating the 32B name as proof that every general-purpose task will improve.
Preparing a document question
The maker’s example separates the document from the question with explicit text delimiters and requests a recognizable final-answer phrase after the reasoning. Follow that structure: provide the relevant document body, keep headings and tables readable, and ask one concrete question. For a report calculation, preserve units, reporting periods and table labels, then request the formula, the values used and a final result. For a policy comparison, label the documents A and B and request a table of matching rules, differences and the supporting section for each row. These are authored applications of the documented numerical and cross-document capabilities.
The example parses reasoning and final-answer text separately. It uses a 10,000-token generation budget, temperature 0.7 and top_p 0.95; those are example settings, not a declared maximum output length. The card also shows an optional identity-setting system message. Preserve this model’s own prompt guidance rather than importing a different checkpoint’s recommendations solely because it appears in the model’s ancestry.
Processing long documents
Tongyi Lab validates context lengths up to 131,072 tokens using YaRN and recommends scaling when combined input and output significantly exceed 32,768 tokens. Its examples describe a factor of four, or a smaller factor such as two for a 65,536-token workload. Static scaling can reduce performance on shorter text, so the card advises against enabling it when ordinary requests stay within 32,768 tokens. The published configuration lists 131,072 positions but no enabled rope_scaling field; configuration and the maker’s conditional long-context procedure should therefore be checked together.
A long input still needs room for reasoning and the answer. Start with the passages needed for the question, remove repeated headers and navigation text, and retain page or section identifiers. Ask for evidence references in the result so a reviewer can trace the conclusion back to the supplied material. A large context allowance helps place more evidence in one request; it does not establish perfect recall of every detail.
What the document training covers
DocQA-RL-1.6K contains 1,600 training problems across three reasoning domains: 600 numerical questions from long specialized documents such as financial reports; 600 logical multiple-choice questions about legal, financial, insurance and production documents; and 400 multi-hop examples that connect evidence across sources. This makes the model a relevant candidate for report arithmetic, rule interpretation and questions requiring several supporting passages. The training domains describe experience with those tasks, not professional certification or guaranteed conclusions.
Practical output and limits
Choose an answer that can be checked: a computed value with units, a rule comparison with section references, or a short conclusion with the chain of supporting facts. If the document omits a required date or definition, ask the model to identify the missing evidence instead of inventing it. The release materials focus on document question answering and do not establish a dedicated tool-calling contract or a separate hard output ceiling. Treat answers as grounded drafts and verify calculations, references and interpretations against the supplied documents.
POST /v1/chat/completionstexttext| Field | Type | Requirement | Values & limits | Default |
|---|---|---|---|---|
Fieldmodel | Typestring | RequirementRequired | Values & limits Tongyi-Zhiwen/QwenLong-L1-32B | 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 $0.00028 / 1K tokens Output · Output tokens $0.0012 / 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": "Tongyi-Zhiwen/QwenLong-L1-32B", "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='Tongyi-Zhiwen/QwenLong-L1-32B',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='Tongyi-Zhiwen/QwenLong-L1-32B',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 Tongyi-Zhiwen/QwenLong-L1-32B
Use the method paths shown above.
Tongyi-Zhiwen/QwenLong-L1-32BQwenLong-L1-32B is Tongyi Lab’s text reasoning model for questions grounded in long documents. It combines a DeepSeek-R1-Distill-Qwen-32B starting point with document-focused reinforcement learning for numerical, logical and cross-document reasoning.
This description covers the model’s capabilities. Available in Ailoot inputs, outputs and options are listed for the selected method below.
Introduction: reasoning over long documents
QwenLong-L1-32B is Tongyi Lab’s long-context text reasoning model. Its purpose is to answer questions whose evidence is spread through a document or across several documents, rather than simply repeat a nearby passage. It is built from DeepSeek-R1-Distill-Qwen-32B and further trained for document-grounded reasoning. Supply text and a question; the model generates a textual reasoning sequence and answer. Scanned pages, audio and video need an appropriate separate conversion step before this text workflow.
How the long-context training works
The QwenLong-L1 framework first warms up a reasoning policy with supervised fine-tuning, then increases context length through a reinforcement-learning curriculum. Difficulty-aware retrospective sampling adjusts which examples are revisited as training progresses. The maker combines rule-based and model-based binary outcome rewards, using group-relative policy optimization to encourage answers that remain grounded in long evidence. This describes the checkpoint’s training approach; it does not mean an individual answer is automatically verified by an external judge.
The released edition
This is the original QwenLong-L1-32B release. The separately named QwenLong-L1-32B-AWQ is an INT4-quantized edition and should retain its own identity. The published configuration for this model uses a Qwen2 causal-language-model architecture with 64 layers and bfloat16 weights. The model’s document specialization is more relevant to selection than treating the 32B name as proof that every general-purpose task will improve.
Preparing a document question
The maker’s example separates the document from the question with explicit text delimiters and requests a recognizable final-answer phrase after the reasoning. Follow that structure: provide the relevant document body, keep headings and tables readable, and ask one concrete question. For a report calculation, preserve units, reporting periods and table labels, then request the formula, the values used and a final result. For a policy comparison, label the documents A and B and request a table of matching rules, differences and the supporting section for each row. These are authored applications of the documented numerical and cross-document capabilities.
The example parses reasoning and final-answer text separately. It uses a 10,000-token generation budget, temperature 0.7 and top_p 0.95; those are example settings, not a declared maximum output length. The card also shows an optional identity-setting system message. Preserve this model’s own prompt guidance rather than importing a different checkpoint’s recommendations solely because it appears in the model’s ancestry.
Processing long documents
Tongyi Lab validates context lengths up to 131,072 tokens using YaRN and recommends scaling when combined input and output significantly exceed 32,768 tokens. Its examples describe a factor of four, or a smaller factor such as two for a 65,536-token workload. Static scaling can reduce performance on shorter text, so the card advises against enabling it when ordinary requests stay within 32,768 tokens. The published configuration lists 131,072 positions but no enabled rope_scaling field; configuration and the maker’s conditional long-context procedure should therefore be checked together.
A long input still needs room for reasoning and the answer. Start with the passages needed for the question, remove repeated headers and navigation text, and retain page or section identifiers. Ask for evidence references in the result so a reviewer can trace the conclusion back to the supplied material. A large context allowance helps place more evidence in one request; it does not establish perfect recall of every detail.
What the document training covers
DocQA-RL-1.6K contains 1,600 training problems across three reasoning domains: 600 numerical questions from long specialized documents such as financial reports; 600 logical multiple-choice questions about legal, financial, insurance and production documents; and 400 multi-hop examples that connect evidence across sources. This makes the model a relevant candidate for report arithmetic, rule interpretation and questions requiring several supporting passages. The training domains describe experience with those tasks, not professional certification or guaranteed conclusions.
Practical output and limits
Choose an answer that can be checked: a computed value with units, a rule comparison with section references, or a short conclusion with the chain of supporting facts. If the document omits a required date or definition, ask the model to identify the missing evidence instead of inventing it. The release materials focus on document question answering and do not establish a dedicated tool-calling contract or a separate hard output ceiling. Treat answers as grounded drafts and verify calculations, references and interpretations against the supplied documents.
POST /v1/messagestextjson, sse| Field | Type | Requirement | Values & limits | Default |
|---|---|---|---|---|
Fieldmodel | Typestring | RequirementRequired | Values & limits Tongyi-Zhiwen/QwenLong-L1-32B | 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 $0.00028 / 1K tokens Output · Output tokens $0.0012 / 1K tokens Input · Cache read $0 / 1K tokens Input · Cache write $0 / 1K tokens Input · Cache write, 1 hour — / 1K tokens |
Ailoot reserves an estimated amount before a request. After the result, it settles the final charge from measured usage. Streaming settlement uses the final usage record. How reservation and settlement work →
{\n "model": "Tongyi-Zhiwen/QwenLong-L1-32B",\n "max_tokens": 128,\n "messages": [\n {\n "role": "user",\n "content": "Summarize the incident report."\n }\n ],\n "stream": false\n}set -eu\n: "${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': 'Tongyi-Zhiwen/QwenLong-L1-32B', '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': 'Tongyi-Zhiwen/QwenLong-L1-32B', '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 Tongyi-Zhiwen/QwenLong-L1-32B
Use the method paths shown above.
Technical example in English
Connect evidence across report sections\n\nDocument section 1 says 240 units were shipped. Section 4 says 30 shipped units were returned before month end. Appendix A says net shipped units equal shipped minus returned units. Calculate net shipped units for the month, cite every section needed and state whether revenue can be calculated from these facts. End with a clearly labelled final answer.A source-linked formula and calculation, a revenue-information limit and a labelled final result.Check that all three source sections support the calculation and that a price or revenue value is not invented.