Open Model WeightsDATALAB · FIELD VERIFIED
Primary sources last checked on 1 Nov 2026. Verification freshness is tracked separately from factual change, and unknown values remain explicit rather than inferred.
CONCEPTUAL MODEL BRIEF · ORIGINAL ILLUSTRATION
A topic-specific editorial visual, not an independent model performance measurement.MODEL INTELLIGENCE · SOURCE-FIRST
Open Model Weights records Surya OCR 2.5 as a 0.689357B-parameter documents and images → ocr, layout and tables open-weight model in the Surya OCR family, with Datalab listed as the developer. The checked source repository provides 1 recognized weight artifacts, format evidence for Safetensors and a verified context window of 262,144 tokens. Values that are not evidenced in the checked sources remain explicitly undisclosed or unknown.
The repository declares Datalab modified OpenRAIL-M model weights. Weight use subject to Datalab modified OpenRAIL-M conditions: publisher permits certain research, personal and startup use under its $5M funding/revenue threshold; broader commercial deployment requires separate terms. The Apache-2.0 code grant does not apply to model weights.. Open Model Weights reports the checked license evidence and classification; this is not legal advice.
No base model is declared in the checked standard lineage metadata, so Open Model Weights does not infer one. Training recipe status: training-described-in-model-card; data disclosure status: described-in-model-card.
+ Save to my shortlistMonitor changes →Local browser monitoring · no background email alerts
DIRECT ANSWER · FEATURED MODEL BRIEFING
Source-grounded editorial layer · reviewed 2026-11-01
Datalab Surya 2.5 is a 689M open-weight OCR and document-understanding checkpoint for text, layout and table structure. Its model weights use a modified OpenRAIL-M license with commercial restrictions; Apache-2.0 applies to code, not automatically to weights. Open Model Weights independently inspected the exact repository revision, listed weight artifacts and declared license; it has not run the model or reproduced publisher benchmark results. The repository includes a separately checked license file.
WHY IT MATTERS
Datalab combines document OCR, reading order, layout classification and table interpretation in a small vision-language checkpoint. The model is aimed at PDFs and scanned pages rather than general outdoor images or ordinary image captioning.
Datalab reports 85.9 on OmniParseBench and 4.5 pages per second on one H100 at 192 DPI under its measured setup. These numbers are informative publisher benchmarks, not independent OMW measurements.
The unified vision-language model can produce layout regions, OCR blocks and table HTML. The manufacturer notes that Surya 2.5 does not provide the row and column geometry supported by a separate earlier Surya 2 checkpoint.
MODEL-SELECTION VIEW
Datalab combines document OCR, reading order, layout classification and table interpretation in a small vision-language checkpoint. The model is aimed at PDFs and scanned pages rather than general outdoor images or ordinary image captioning.
Full-page outputs and structured HTML tables have different constraints from line detection. The 2.5 inference package also relies on separately installed server backends, so downloading one weight file does not deliver an entire document processing service.
The Surya Python code is Apache-2.0, but Datalab explicitly gives the model weights a modified OpenRAIL-M license. The publisher describes free use for research, personal contexts and startups below its stated five-million-dollar funding/revenue threshold.
Larger businesses and other commercial deployments may require a separate paid license. The precise definitions and limits belong to the weight terms and must be reviewed for the deployment; no unrestricted commercial permission may be inferred from the code license.
Datalab reports 85.9 on OmniParseBench and 4.5 pages per second on one H100 at 192 DPI under its measured setup. These numbers are informative publisher benchmarks, not independent OMW measurements.
Accuracy will vary with page rotation, language, handwriting, columns, scans and forms. Hardware throughput also depends on batch size, preprocessing, inference engine and page dimensions, so a reported H100 benchmark is not a general deployment guarantee.
The unified vision-language model can produce layout regions, OCR blocks and table HTML. The manufacturer notes that Surya 2.5 does not provide the row and column geometry supported by a separate earlier Surya 2 checkpoint.
Version changes affect output schema and the way existing pipelines interpret coordinates, HTML and table information. Migration should validate downstream parsers instead of assuming old and new API objects are interchangeable.
A 689M open-weight document specialist with explicit distribution and task scope is valuable for auditability. Its Notable curation recognizes specialist significance without projecting publisher scores into independent leadership.
Leading would require a repeatable, external document benchmark with matched data rights, resolution and exact model revision. Until that is available, the profile separates source-backed performance statements from independently replicated outcomes.
Pin exact weight SHA, the Surya package version, vLLM or llama.cpp backend and page preprocessing settings. Start with representative PDFs containing tables, languages and failure cases; review normalized text and structured outputs side by side.
Before processing real customer documents, confirm rights to the model weights, document privacy controls, storage retention and error handling. Report CER, layout fidelity and table extraction quality separately from pages-per-second throughput.
IMPLEMENTATION & EVALUATION
Reproducible operator checklist; these steps are not claims of an OMW model run.
A named model does not guarantee identical checkpoint revisions.
Record 40-character SHA, original file names, configuration and tokenizer/processor versions.
Unpinned revisions can silently change weights or usage requirements.
A specialist checkpoint can fail outside the publisher's demonstrated conditions.
Document actual prompt templates, modality preprocessing, inference backend and compatible software versions.
Using unsupported serving backends can invalidate comparisons.
Model card labels and code licenses are not automatically permission for every deployed use.
Check exact weight terms, notices, tool permissions, data privacy and runtime isolation.
Restricted weight rights or unsafe action tools can make a deployment unsuitable.
Vendor benchmark figures are useful evidence, but are not OMW-run observations.
Pre-register held-out tasks and compare baseline checkpoint, hardware, quality, memory and error rates.
Attributing first-party results to independent inference would mislead readers.
Proposed tests, not OMW benchmark results. Read the official model card at the pinned source revision ↗ Inspect OMW field evidence ↗
PRIMARY & INDEPENDENT SOURCES
FAQ
Yes, the developer describes the main OCR/document vision-language checkpoint as approximately 689 million parameters; companion detection models are separate components.
No. The Python code is Apache-2.0, while the downloaded model weights use a modified OpenRAIL-M license with conditional commercial permissions.
No. Datalab states a threshold for certain free commercial/startup use and requires broader commercial licensing for other cases; consult the exact current terms.
No. Those are Datalab-published metrics for named tests and conditions; Open Model Weights has not rerun them.
Not in the published 2.5 table mode. The manufacturer directs users requiring that specific geometry to its earlier Surya 2 checkpoint.
The official Surya package describes a vLLM backend for NVIDIA GPU or llama.cpp server for CPU and Apple Silicon, with separate installation requirements.
03 · PRACTICAL TECHNICAL PROFILE
Inspect weight-memory estimates, model lineage and runtime evidence separately from promotional or untested performance claims.
DEPLOYMENT PROFILE
These bars visualize weight-only decimal-GB estimates. They exclude KV cache, activations, runtime overhead, optimizer state and sharding.
LICENSE & USE
Weight use subject to Datalab modified OpenRAIL-M conditions: publisher permits certain research, personal and startup use under its $5M funding/revenue threshold; broader commercial deployment requires separate terms. The Apache-2.0 code grant does not apply to model weights.
This section summarizes the checked repository evidence. It is not legal advice. Inspect license evidence ↗
LINEAGE & PROVENANCE
Open Model Weights does not infer a hidden base model or training corpus when the checked sources do not declare one.
RUNTIME SIGNALS
Declared and model-card-mentioned runtimes are discovery signals. They are not treated as Open Model Weights runtime tests unless explicitly measured.
EVIDENCE TRAIL
This is where the article remains auditable: source links on one side, the registry's observed change history on the other.
Checked Hugging Face API metadata, exact repository file list, model card and config when accessible.
04 · EVIDENCE AND UPDATES
Inspect dated source checks and any differences between published claims and the evidence; these checks are not independent model benchmarks.
SOURCE-BASED VISUAL BRIEF · NOTABLE
Three evidence-linked visual summaries: parameter scale, illustrative precision memory and verifiable provenance. No performance ranking is implied.
Active parameter count not established.
Repository-derived or publisher-described parameter countFRESHNESS & VERIFICATION
A verification check does not pretend the model changed. Open Model Weights updates the verification date after a real primary-source check, while factual-update dates move only when retained source evidence changes.
CONFLICT INTELLIGENCE
05 · SOURCE RECORD
Full repository-level attributes, licensing, format details and provenance are available below without hiding or rewriting the original data.
01 · WEIGHTS
02 · LICENSE
Classification describes the checked license terms; it is not legal advice.
License / model-card evidence ↗03 · HARDWARE
Weight-only decimal-GB estimate from total parameter count; excludes KV cache, activations, runtime overhead, optimizer state and sharding.
04 · FORMATS
“Not observed” means no matching repository artifact/name/dtype signal was found; it does not prove no third-party conversion exists.
05 · LINEAGE
06 · TRAINING ASSETS
07 · RUNTIME SUPPORT
Open Model Weights separates model-repository evidence from external runtime-release evidence. An external support statement does not change the official model repository's weight formats and does not imply an Open Model Weights inference benchmark.
08 · SOURCES
09 · VERIFICATION HISTORY
Checked Hugging Face API metadata, exact repository file list, model card and config when accessible.
MODELS FOR AI AGENTS · ROLE EVIDENCE
Document OCR & layout. A supporting component. This checkpoint does not, by category alone, serve as the planning and tool-execution controller.
The curated notable tier is not an agent-readiness rating. Model-specific tool-call success, long-horizon task accuracy and compatible runtime configurations are not independently established here.
Source-linked license: Datalab modified OpenRAIL-M model weights (allowed-with-conditions commercial-use classification; see original terms).
Context: 262144 configured tokens.
BF16 weights: 1.38 estimated decimal GB, not runtime VRAM.
Compare agent model candidates → · Agent evaluation method → · Inspect original model repository ↗
These records are separately sourced. They do not replace independent model-plus-runtime tests.
No model-specific agent capability evaluation is recorded for this checkpoint. Missing evidence is unknown, not a failed task.
Read evidence details and benchmarks → · Use the agent model selector →