Chandra Ocr 2 — Hardware Requirements & GPU Compatibility
Vision
Chandra OCR 2 is Datalab's 5.3-billion-parameter vision-language model for OCR and document conversion, turning scanned pages and PDFs into markdown, HTML, or JSON while preserving layout. It runs on a Qwen3.5-based backbone mixing linear attention with periodic full-attention layers, and handles handwriting, forms, tables, and math across 90-plus languages. It is the second generation of Datalab's Chandra model, improved over the first release. At just over 5 billion parameters, it runs on a single consumer GPU once quantized. Context extends to 262,144 tokens, generous for multi-page documents. It carries a modified OpenRAIL-M license: free for research, personal use, and startups under $2 million in funding or revenue, but it bars competing against Datalab's own API, more restrictive than a standard permissive license. Published in March 2026, following the original Chandra.
2.5M downloads
518 likesSpecifications
- Publisher
- datalab-to
- Parameters
- 5.3B
- Architecture
- Qwen3_5ForConditionalGeneration
- Context Length
- 262,144 tokens
- Vocabulary Size
- 248,320
- Release Date
- 2026-03-16
- License
- openrail
Get Started
HuggingFace
How Much VRAM Does Chandra Ocr 2 Need?
Select a quantization to see compatible GPUs below.
| Quantization | Bits | VRAM | + Context | File Size | Quality |
|---|---|---|---|---|---|
| Q2_K | 3.40 | 2.7 GB | 24.0 GB | 2.25 GB | 2-bit quantization with K-quant improvements |
| Q3_K_S | 3.50 | 2.8 GB | 24.1 GB | 2.32 GB | 3-bit small quantization |
| Q3_K_M | 3.90 | 3.0 GB | 24.4 GB | 2.58 GB | 3-bit medium quantization |
| Q4_0 | 4.00 | 3.1 GB | 24.4 GB | 2.65 GB | 4-bit legacy quantization |
| Q4_K_M | 4.80 | 3.6 GB | 24.9 GB | 3.18 GB | 4-bit medium quantization — most popular sweet spot |
| Q5_K_M | 5.70 | 4.2 GB | 25.6 GB | 3.77 GB | 5-bit medium quantization — good quality/size tradeoff |
| Q6_K | 6.60 | 4.8 GB | 26.1 GB | 4.37 GB | 6-bit quantization, very good quality |
| Q8_0 | 8.00 | 5.8 GB | 27.1 GB | 5.30 GB | 8-bit quantization, near-lossless |
Which GPUs Can Run Chandra Ocr 2?
Q4_K_M · 3.6 GBChandra Ocr 2 (Q4_K_M) requires 3.6 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 5+ GB is recommended. Using the full 262K context window can add up to 21.3 GB, bringing total usage to 24.9 GB. 52 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti.
Runs great
— Plenty of headroomWhich Devices Can Run Chandra Ocr 2?
Q4_K_M · 3.6 GB59 devices with unified memory can run Chandra Ocr 2, including NVIDIA DGX H100, NVIDIA DGX A100 640GB, iPhone 17.
Runs great
— Plenty of headroom
Decent
— Enough memory, may be tightWhere to Download Chandra Ocr 2
Community quantizations of this model — GGUF for llama.cpp, Ollama, and LM Studio, plus AWQ/MLX variants where available.
Frequently Asked Questions
- How much VRAM does Chandra Ocr 2 need?
Chandra Ocr 2 requires 3.6 GB of VRAM at Q4_K_M, or 11.1 GB at BF16. Full 262K context adds up to 21.3 GB (24.9 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 5.3B × 4.8 bits ÷ 8 = 3.2 GB
KV Cache + Overhead ≈ 0.4 GB (at 2K context + ~0.3 GB framework)
KV Cache + Overhead ≈ 21.8 GB (at full 262K context)
VRAM usage by quantization
Q4_K_M3.6 GBQ4_K_M + full context24.9 GB- What's the best quantization for Chandra Ocr 2?
For Chandra Ocr 2, Q4_K_M (3.6 GB) offers the best balance of quality and VRAM usage. Q5_0 (3.8 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 2.7 GB.
VRAM requirement by quantization
Q2_K2.7 GBQ4_03.1 GBQ4_K_M ★3.6 GBQ5_03.8 GBQ5_K_M4.2 GBBF1611.1 GB★ Recommended — best balance of quality and VRAM usage.
- Can I run Chandra Ocr 2 on a Mac?
Chandra Ocr 2 requires at least 2.7 GB at Q2_K, which exceeds the unified memory of most consumer Macs. You would need a Mac Studio or Mac Pro with a high-memory configuration.
- Can I run Chandra Ocr 2 locally?
Yes — Chandra Ocr 2 can run locally on consumer hardware. At Q4_K_M quantization it needs 3.6 GB of VRAM. Popular tools include Ollama, LM Studio, and llama.cpp.
- How fast is Chandra Ocr 2?
At Q4_K_M, Chandra Ocr 2 can reach ~1315 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~180 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = (bandwidth GB/s ÷ model GB) × efficiency
Example: NVIDIA B200 → 8000 ÷ 3.6 × 0.65 = ~1425 tok/s
Estimated speed at Q4_K_M (3.6 GB)
~1425 tok/s~180 tok/s~1425 tok/s~1315 tok/sReal-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
- What's the download size of Chandra Ocr 2?
At Q4_K_M, the download is about 3.18 GB. The full-precision BF16 version is 10.59 GB. The smallest option (Q2_K) is 2.25 GB.
- Which GPUs can run Chandra Ocr 2?
52 consumer GPUs can run Chandra Ocr 2 at Q4_K_M (3.6 GB). Top options include AMD Radeon RX 6700 XT, AMD Radeon RX 6800, AMD Radeon RX 6800 XT. 52 GPUs have plenty of headroom for comfortable inference.
- Which devices can run Chandra Ocr 2?
59 devices with unified memory can run Chandra Ocr 2 at Q4_K_M (3.6 GB), including AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10, Apple iPhone 17 Pro, Asus ROG Flow Z13 (2025, Ryzen AI Max+ 395, 128 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.