Skip to content

A fast, helpful, and open-source document parser

docs(suryaocr): accurate language stats and document required inference backend

Fu Dai275e93523e1c
diff --git a/ocr/README.md b/ocr/README.md
index 3bb5246..e2ef24d 100644
--- a/ocr/README.md
+++ b/ocr/README.md
@@ -12,7 +12,7 @@ These servers allow you to use alternative OCR engines instead of the built-in T
| Speed | Moderate | Moderate | Fast (2-3x) | Moderate (GPU recommended) |
| Accuracy (Latin) | Good | Good | Good | Excellent |
| Accuracy (CJK) | Fair | Good | Excellent | Excellent |
-| Languages | 100+ | 80+ | 80+ | 90+ |
+| Languages | 100+ | 80+ | 80+ | 91 |
| Memory | In-process | Separate | Separate | Separate |
**Recommendations:**
diff --git a/ocr/suryaocr/.dockerignore b/ocr/suryaocr/.dockerignore
new file mode 100644
index 0000000..213b2cb
--- /dev/null
+++ b/ocr/suryaocr/.dockerignore
@@ -0,0 +1,3 @@
+.venv
+__pycache__
+.pytest_cache
diff --git a/ocr/suryaocr/README.md b/ocr/suryaocr/README.md
index 5e59ca3..83e9a20 100644
--- a/ocr/suryaocr/README.md
+++ b/ocr/suryaocr/README.md
@@ -16,6 +16,24 @@ uv run server.py
The first run downloads Surya model weights from Hugging Face and may take a few
minutes; weights are cached afterward.
+## Inference backend (required)
+
+Surya 2 is a VLM-backed model: text recognition runs through a separate
+inference backend, which you must provide. Surya does **not** bundle it.
+
+- **CPU / local (llama.cpp):** install the `llama-server` binary so Surya's
+ `llamacpp` backend can spawn it:
+ - macOS: `brew install llama.cpp`
+ - Linux: `brew install llama.cpp`, or download a release from
+ https://github.com/ggml-org/llama.cpp/releases and put `llama-server` on
+ your `PATH` (or set `LLAMA_CPP_BINARY=/path/to/llama-server`).
+- **GPU (vllm):** set `SURYA_INFERENCE_BACKEND=vllm` (requires a CUDA GPU).
+- **External server:** point `SURYA_INFERENCE_URL` at an already-running Surya
+ inference server to attach without spawning one locally.
+
+Without a backend, startup succeeds and `GET /health` works, but `POST /ocr`
+returns a 500 with "llama-server binary not found".
+
## Usage
The service exposes:
@@ -55,10 +73,16 @@ block's HTML stripped to plain text. This conforms to the LiteParse OCR API spec
## Supported Languages
-Surya 2 supports 90+ languages with a single multilingual model, with strong
-performance on non-Latin scripts. No `language` parameter is required.
+Surya 2 is a single multilingual model — no `language` parameter is required
+(the `language` field is accepted but ignored).
+
+Per Surya's own benchmark, it scores an **87.2% overall pass rate across 91
+languages**, with 38 of the 91 languages scoring ≥ 90% and 76 scoring ≥ 80%,
+covering text accuracy, layout, tables, math, and reading order. It has strong
+performance across both Latin and non-Latin scripts.
-Full details: https://github.com/datalab-to/surya
+- Full 91-language results: https://github.com/datalab-to/surya/blob/master/static/docs/multilingual.md
+- Project overview: https://github.com/datalab-to/surya
## Device / GPU