Gemma 3 12B IT is Google’s instruction-tuned 12B-size vision-language model. It belongs to the Gemma 3 family, which combines text generation with image understanding and multilingual training. The IT suffix selects the instruction-following checkpoint, rather than the pre-trained model. Its result is text: an explanation, summary, answer, draft or code derived from the written and visual material in a request.
The 12B model has the family’s 128K-token context for both the input and generated response. The output uses the portion remaining after the request; the card does not grant a second independent 128K output window. Base image views are normalized to 896×896 and represented by 256 visual tokens each. When preparing a collection of pages, retain the page identifiers and allocate space for the result. Ask a specific question about the relevant content rather than relying on a large window to select the right issue automatically.
The maker reports 12 trillion pre-training tokens for this size. Training material includes multilingual web documents, code, mathematics and images, with coverage of more than 140 languages in the family. The report describes knowledge distillation and an improved instruction-tuning recipe with reinforcement learning. Those stages target helpfulness, mathematical and code reasoning, instruction following and multilingual interaction. A distinct training budget and checkpoint identity do not alone establish the quality of a particular answer, so compare the output with the source task.
The architecture uses a decoder-only transformer with grouped-query attention and a repeating pattern of five local-attention layers to one global layer. Local layers use a 1,024-token span, while global attention supports the longer context. A SigLIP vision encoder, shared by the 4B, 12B and 27B models, converts images into representations consumed by the language model. The SentencePiece tokenizer preserves whitespace, splits digits and includes byte-level encoding. This is relevant when code formatting, numerical notation or multilingual text needs to remain intact.
The vision workflow also includes Pan and Scan: adaptive non-overlapping crops help retain content from wide or high-resolution sources before each view reaches the fixed-resolution encoder. Additional views change the material being processed. For a technical document, give a readable page and, where necessary, a clear view of the table or diagram being questioned. Preserve the surrounding definition of a unit, symbol or column so a local visual reading does not lose its meaning.
The native IT prompt format uses user and model turns and does not have a separate system role. Place the desired role, response rules and task in the initial user prompt, using the designated formatter or processor for the actual conversation. An original document-analysis request might compare two procedure descriptions, return a table of matching requirements and identify any conflict. Label the sources, state which requirement has priority and ask the model to separate the supplied facts from its interpretation.
The intended uses in the maker’s card include summarization, conversational assistance, image-data extraction, writing and learning. For a multilingual explanation, give the target reader, terminology and level of detail. For an image question, identify the objects or relations to discuss and request observations supporting the final answer. For code, include the actual function and expected behavior; the result can propose an implementation, while a separate application must run and check it. The report publishes quantized family variants separately, so this 12B IT identity should not be silently replaced by a different quantized checkpoint when recording results.
The maker identifies limits from training coverage, factual inaccuracies, language nuance and complex open-ended tasks. Google also cautions that a plausible language pattern can produce an answer that misses an everyday real-world constraint. Small visual details can be lost or misread, and long inputs still need clear priorities. Retain the original pages, verify extracted labels and quantities, and validate any requested structured-looking text before using it as data. Evaluate the exact 12B IT model on representative questions and languages. A useful final artifact states the source-based answer, the reasoning needed to understand it and the remaining uncertainty, without pretending that a generated explanation is a completed tool action or a live knowledge lookup.
", "aboutText": "Gemma 3 12B IT is Google’s instruction-tuned 12B-size vision-language model. It belongs to the Gemma 3 family, which combines text generation with image understanding and multilingual training. The IT suffix selects the instruction-following checkpoint, rather than the pre-trained model. Its result is text: an explanation, summary, answer, draft or code derived from the written and visual material in a request.\n\nThe 12B model has the family’s 128K-token context for both the input and generated response. The output uses the portion remaining after the request; the card does not grant a second independent 128K output window. Base image views are normalized to 896×896 and represented by 256 visual tokens each. When preparing a collection of pages, retain the page identifiers and allocate space for the result. Ask a specific question about the relevant content rather than relying on a large window to select the right issue automatically.\n\nThe maker reports 12 trillion pre-training tokens for this size. Training material includes multilingual web documents, code, mathematics and images, with coverage of more than 140 languages in the family. The report describes knowledge distillation and an improved instruction-tuning recipe with reinforcement learning. Those stages target helpfulness, mathematical and code reasoning, instruction following and multilingual interaction. A distinct training budget and checkpoint identity do not alone establish the quality of a particular answer, so compare the output with the source task.\n\nThe architecture uses a decoder-only transformer with grouped-query attention and a repeating pattern of five local-attention layers to one global layer. Local layers use a 1,024-token span, while global attention supports the longer context. A SigLIP vision encoder, shared by the 4B, 12B and 27B models, converts images into representations consumed by the language model. The SentencePiece tokenizer preserves whitespace, splits digits and includes byte-level encoding. This is relevant when code formatting, numerical notation or multilingual text needs to remain intact.\n\nThe vision workflow also includes Pan and Scan: adaptive non-overlapping crops help retain content from wide or high-resolution sources before each view reaches the fixed-resolution encoder. Additional views change the material being processed. For a technical document, give a readable page and, where necessary, a clear view of the table or diagram being questioned. Preserve the surrounding definition of a unit, symbol or column so a local visual reading does not lose its meaning.\n\nThe native IT prompt format uses user and model turns and does not have a separate system role. Place the desired role, response rules and task in the initial user prompt, using the designated formatter or processor for the actual conversation. An original document-analysis request might compare two procedure descriptions, return a table of matching requirements and identify any conflict. Label the sources, state which requirement has priority and ask the model to separate the supplied facts from its interpretation.\n\nThe intended uses in the maker’s card include summarization, conversational assistance, image-data extraction, writing and learning. For a multilingual explanation, give the target reader, terminology and level of detail. For an image question, identify the objects or relations to discuss and request observations supporting the final answer. For code, include the actual function and expected behavior; the result can propose an implementation, while a separate application must run and check it. The report publishes quantized family variants separately, so this 12B IT identity should not be silently replaced by a different quantized checkpoint when recording results.\n\nThe maker identifies limits from training coverage, factual inaccuracies, language nuance and complex open-ended tasks. Google also cautions that a plausible language pattern can produce an answer that misses an everyday real-world constraint. Small visual details can be lost or misread, and long inputs still need clear priorities. Retain the original pages, verify extracted labels and quantities, and validate any requested structured-looking text before using it as data. Evaluate the exact 12B IT model on representative questions and languages. A useful final artifact states the source-based answer, the reasoning needed to understand it and the remaining uncertainty, without pretending that a generated explanation is a completed tool action or a live knowledge lookup.", "editorialOverlay": { "copy_fields": { "practical_html": "Technical example in English
Explain an experiment to a learner\n\nExplain this class experiment to a 12-year-old in plain English. Two identical cups contain 200 ml of water at the same starting temperature. One is wrapped in wool; the other is uncovered. After 20 minutes the wrapped cup is warmer. Give the conclusion, what was controlled, and one improvement. Do not invent temperatures.Three short labelled parts: conclusion, controlled conditions and one practical improvement.The explanation distinguishes slowing heat loss from creating heat and preserves the given quantities without new measurements.google/gemma-3-12b-itGemma 3 12B IT is a separate instruction-tuned Gemma 3 edition for text generation and image understanding. Use it for document questions, explanations and multilingual writing with a shared 128K context, keeping the 12B checkpoint and its native conversation format explicit.
This description covers the model’s capabilities. Available in Ailoot inputs, outputs and options are listed for the selected method below.
Gemma 3 12B IT is Google’s instruction-tuned 12B-size vision-language model. It belongs to the Gemma 3 family, which combines text generation with image understanding and multilingual training. The IT suffix selects the instruction-following checkpoint, rather than the pre-trained model. Its result is text: an explanation, summary, answer, draft or code derived from the written and visual material in a request.
The 12B model has the family’s 128K-token context for both the input and generated response. The output uses the portion remaining after the request; the card does not grant a second independent 128K output window. Base image views are normalized to 896×896 and represented by 256 visual tokens each. When preparing a collection of pages, retain the page identifiers and allocate space for the result. Ask a specific question about the relevant content rather than relying on a large window to select the right issue automatically.
The maker reports 12 trillion pre-training tokens for this size. Training material includes multilingual web documents, code, mathematics and images, with coverage of more than 140 languages in the family. The report describes knowledge distillation and an improved instruction-tuning recipe with reinforcement learning. Those stages target helpfulness, mathematical and code reasoning, instruction following and multilingual interaction. A distinct training budget and checkpoint identity do not alone establish the quality of a particular answer, so compare the output with the source task.
The architecture uses a decoder-only transformer with grouped-query attention and a repeating pattern of five local-attention layers to one global layer. Local layers use a 1,024-token span, while global attention supports the longer context. A SigLIP vision encoder, shared by the 4B, 12B and 27B models, converts images into representations consumed by the language model. The SentencePiece tokenizer preserves whitespace, splits digits and includes byte-level encoding. This is relevant when code formatting, numerical notation or multilingual text needs to remain intact.
The vision workflow also includes Pan and Scan: adaptive non-overlapping crops help retain content from wide or high-resolution sources before each view reaches the fixed-resolution encoder. Additional views change the material being processed. For a technical document, give a readable page and, where necessary, a clear view of the table or diagram being questioned. Preserve the surrounding definition of a unit, symbol or column so a local visual reading does not lose its meaning.
The native IT prompt format uses user and model turns and does not have a separate system role. Place the desired role, response rules and task in the initial user prompt, using the designated formatter or processor for the actual conversation. An original document-analysis request might compare two procedure descriptions, return a table of matching requirements and identify any conflict. Label the sources, state which requirement has priority and ask the model to separate the supplied facts from its interpretation.
The intended uses in the maker’s card include summarization, conversational assistance, image-data extraction, writing and learning. For a multilingual explanation, give the target reader, terminology and level of detail. For an image question, identify the objects or relations to discuss and request observations supporting the final answer. For code, include the actual function and expected behavior; the result can propose an implementation, while a separate application must run and check it. The report publishes quantized family variants separately, so this 12B IT identity should not be silently replaced by a different quantized checkpoint when recording results.
The maker identifies limits from training coverage, factual inaccuracies, language nuance and complex open-ended tasks. Google also cautions that a plausible language pattern can produce an answer that misses an everyday real-world constraint. Small visual details can be lost or misread, and long inputs still need clear priorities. Retain the original pages, verify extracted labels and quantities, and validate any requested structured-looking text before using it as data. Evaluate the exact 12B IT model on representative questions and languages. A useful final artifact states the source-based answer, the reasoning needed to understand it and the remaining uncertainty, without pretending that a generated explanation is a completed tool action or a live knowledge lookup.
POST /v1/chat/completionstexttext| Field | Type | Requirement | Values & limits | Default |
|---|---|---|---|---|
Fieldmodel | Typestring | RequirementRequired | Values & limits google/gemma-3-12b-it | 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.
Match each charge to the conditions shown. Use one rate for each set of alternative configurations.
| Configuration | Rate schedule | Context tokens, more than | Context tokens, up to and including | Charge |
|---|---|---|---|---|
| ConfigurationBase token rates | Rate scheduleBase rate | Context tokens, more than— | Context tokens, up to and includingNo upper bound | Charge Input · Input tokens $0.0001 / 1K tokens Output · Output tokens $0.0003 / 1K tokens Input · Cache read — / 1K tokens Input · Cache write — / 1K tokens Input · Cache write, 1 hour — / 1K tokens |
| ConfigurationRequest-length tier 1 | Rate schedule— | Context tokens, more than0 | Context tokens, up to and includingNo upper bound | Charge Input · Input tokens $0.0001 / 1K tokens Output · Output tokens $0.0003 / 1K tokens Input · Cache read — / 1K tokens Input · Cache write — / 1K tokens Input · Cache write, 1 hour — / 1K tokens |
Ailoot reserves an estimated amount before a request. After the result, it settles the final charge from measured usage. Streaming settlement uses the final usage record. How reservation and settlement work →
set -eu\n: "${AILOOT_API_KEY:?Set your Ailoot API key}"\npython3 - <<'PY_REQUEST'\nimport json,os\nfrom pathlib import Path\np=json.loads('{"model": "google/gemma-3-12b-it", "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='google/gemma-3-12b-it',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='google/gemma-3-12b-it',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 google/gemma-3-12b-it
Use the method paths shown above.
google/gemma-3-12b-itGemma 3 12B IT is a separate instruction-tuned Gemma 3 edition for text generation and image understanding. Use it for document questions, explanations and multilingual writing with a shared 128K context, keeping the 12B checkpoint and its native conversation format explicit.
This description covers the model’s capabilities. Available in Ailoot inputs, outputs and options are listed for the selected method below.
Gemma 3 12B IT is Google’s instruction-tuned 12B-size vision-language model. It belongs to the Gemma 3 family, which combines text generation with image understanding and multilingual training. The IT suffix selects the instruction-following checkpoint, rather than the pre-trained model. Its result is text: an explanation, summary, answer, draft or code derived from the written and visual material in a request.
The 12B model has the family’s 128K-token context for both the input and generated response. The output uses the portion remaining after the request; the card does not grant a second independent 128K output window. Base image views are normalized to 896×896 and represented by 256 visual tokens each. When preparing a collection of pages, retain the page identifiers and allocate space for the result. Ask a specific question about the relevant content rather than relying on a large window to select the right issue automatically.
The maker reports 12 trillion pre-training tokens for this size. Training material includes multilingual web documents, code, mathematics and images, with coverage of more than 140 languages in the family. The report describes knowledge distillation and an improved instruction-tuning recipe with reinforcement learning. Those stages target helpfulness, mathematical and code reasoning, instruction following and multilingual interaction. A distinct training budget and checkpoint identity do not alone establish the quality of a particular answer, so compare the output with the source task.
The architecture uses a decoder-only transformer with grouped-query attention and a repeating pattern of five local-attention layers to one global layer. Local layers use a 1,024-token span, while global attention supports the longer context. A SigLIP vision encoder, shared by the 4B, 12B and 27B models, converts images into representations consumed by the language model. The SentencePiece tokenizer preserves whitespace, splits digits and includes byte-level encoding. This is relevant when code formatting, numerical notation or multilingual text needs to remain intact.
The vision workflow also includes Pan and Scan: adaptive non-overlapping crops help retain content from wide or high-resolution sources before each view reaches the fixed-resolution encoder. Additional views change the material being processed. For a technical document, give a readable page and, where necessary, a clear view of the table or diagram being questioned. Preserve the surrounding definition of a unit, symbol or column so a local visual reading does not lose its meaning.
The native IT prompt format uses user and model turns and does not have a separate system role. Place the desired role, response rules and task in the initial user prompt, using the designated formatter or processor for the actual conversation. An original document-analysis request might compare two procedure descriptions, return a table of matching requirements and identify any conflict. Label the sources, state which requirement has priority and ask the model to separate the supplied facts from its interpretation.
The intended uses in the maker’s card include summarization, conversational assistance, image-data extraction, writing and learning. For a multilingual explanation, give the target reader, terminology and level of detail. For an image question, identify the objects or relations to discuss and request observations supporting the final answer. For code, include the actual function and expected behavior; the result can propose an implementation, while a separate application must run and check it. The report publishes quantized family variants separately, so this 12B IT identity should not be silently replaced by a different quantized checkpoint when recording results.
The maker identifies limits from training coverage, factual inaccuracies, language nuance and complex open-ended tasks. Google also cautions that a plausible language pattern can produce an answer that misses an everyday real-world constraint. Small visual details can be lost or misread, and long inputs still need clear priorities. Retain the original pages, verify extracted labels and quantities, and validate any requested structured-looking text before using it as data. Evaluate the exact 12B IT model on representative questions and languages. A useful final artifact states the source-based answer, the reasoning needed to understand it and the remaining uncertainty, without pretending that a generated explanation is a completed tool action or a live knowledge lookup.
POST /v1/messagestextjson, sse| Field | Type | Requirement | Values & limits | Default |
|---|---|---|---|---|
Fieldmodel | Typestring | RequirementRequired | Values & limits google/gemma-3-12b-it | 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.
Match each charge to the conditions shown. Use one rate for each set of alternative configurations.
| Configuration | Rate schedule | Context tokens, more than | Context tokens, up to and including | Charge |
|---|---|---|---|---|
| ConfigurationBase token rates | Rate scheduleBase rate | Context tokens, more than— | Context tokens, up to and includingNo upper bound | Charge Input · Input tokens $0.0001 / 1K tokens Output · Output tokens $0.0003 / 1K tokens Input · Cache read — / 1K tokens Input · Cache write — / 1K tokens Input · Cache write, 1 hour — / 1K tokens |
| ConfigurationRequest-length tier 1 | Rate schedule— | Context tokens, more than0 | Context tokens, up to and includingNo upper bound | Charge Input · Input tokens $0.0001 / 1K tokens Output · Output tokens $0.0003 / 1K tokens Input · Cache read — / 1K tokens Input · Cache write — / 1K tokens Input · Cache write, 1 hour — / 1K tokens |
Ailoot reserves an estimated amount before a request. After the result, it settles the final charge from measured usage. Streaming settlement uses the final usage record. How reservation and settlement work →
{\n "model": "google/gemma-3-12b-it",\n "max_tokens": 128,\n "messages": [\n {\n "role": "user",\n "content": "Summarize the incident report."\n }\n ],\n "stream": false\n}set -eu\n: "${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': 'google/gemma-3-12b-it', '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': 'google/gemma-3-12b-it', '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 google/gemma-3-12b-it
Use the method paths shown above.
Technical example in English
Explain an experiment to a learner\n\nExplain this class experiment to a 12-year-old in plain English. Two identical cups contain 200 ml of water at the same starting temperature. One is wrapped in wool; the other is uncovered. After 20 minutes the wrapped cup is warmer. Give the conclusion, what was controlled, and one improvement. Do not invent temperatures.Three short labelled parts: conclusion, controlled conditions and one practical improvement.The explanation distinguishes slowing heat loss from creating heat and preserves the given quantities without new measurements.