High-capacity 32B multilingual vision model delivering deep visual reasoning, diagram analysis, and rich document understanding across 23 languages.
Technical Specifications
- Provider
- Cohere
- Category
- open-source
- Release Date
- 2025-03
- Batch API Discount
- 50% off async batch
- Consumer Subscription
- Free / Self-Hosted
- Reasoning Effort Control
- Fixed
- Structured Outputs
- Yes
- Output Modalities
- text
- Official Documentation
- Visit Official Docs ↗
Data Privacy, Licensing & Deployment Governance
Official Provider Release Intelligence, API Quotas & Agentic Specs
- Official API Model ID (SDK String)
- aya-vision-32b
- Supported SDK & API Protocol
- OpenAI-Compatible REST / vLLM / SGLang / HuggingFace TGI
- Lifecycle & Deprecation Status
- Active GA (Guaranteed support through 2027+)
- Production Rate Limits (RPM & TPM)
- Unlimited (Hardware-bound on self-hosted GPU) / 10,000 RPM Cloud · Unlimited local throughput / 10M+ TPM Managed Cloud
- Long-Context Pricing Tier (>128K / >200K)
- Flat rate ($0.35/1M in) across full 16K window
- Prompt Cache Write & TTL Policy
- Cached input at $0.08/1M tokens
- Computer Use & Browser GUI Agent
- API / CLI Tool-Use & DOM Agent ready (no raw pixel desktop click loop)
- MCP & Parallel Tool Calling
- Native MCP Client/Server ready · Parallel function calling · 100% Strict JSON Schema
- Web Search Grounding & Code Sandbox
- Supported via external Tavily/Exa Search & E2B/Docker Sandbox tools
- Fine-Tuning & Distillation Support
- Full Weights SFT, LoRA, QLoRA, DPO & GRPO RL (Unsloth / Axolotl / Llama-Factory)
- Multimodal Tokenization Billing
- Text BPE (~4 chars/token) · Vision (~258–765 tokens per image tile)
- Weights Precision & Quantization
- BF16 / FP16, FP8 Dynamic, INT4 AWQ / GPTQ, GGUF (Q4_K_M / Q8_0)
- Minimum GPU VRAM Requirement
- 20 GB VRAM (4-bit Quantized) · 69 GB VRAM (Native BF16)
- Supported Inference Engines
- vLLM, SGLang, Ollama, llama.cpp, TensorRT-LLM, HuggingFace TGI
- System Card & Safety Framework
- Open-Weights System Card + Customizable Guardrail Layer (Llama Guard / ShieldGemma)
- Enterprise Compliance (SOC2 / HIPAA / ZDR)
- 100% Air-Gapped On-Premises / Private VPC · Full HIPAA, GDPR & SOC2 Control
Aya Vision 32B Cost, Context, Hardware & API Integration
For single-user chat, a $20/mo subscription caps your bill below this API volume.
Break-even point: ~402 requests/day equals $20/month.
Same Monthly Workload Cost Across Popular Reference Models
Live Prompt Paste Tokenizer & Cost Counter
Paste real system instructions, JSON schemas, or code to measure BPE token counts across tokenizers.
Hybrid Prompt Routing Cascade Simulator
Route routine queries to a fast lightweight model ($0.15/1M) and escalate only complex reasoning turns to Aya Vision 32B:
Benchmark Scores & 6-Axis Capability Radar
- SWE-bench Verified
- 32%
- GPQA Diamond
- 51.5%
- MMLU-Pro
- 74%
- HumanEval
- 78%
- Chatbot Arena Elo
- 1280pts
Strengths
- Deep multilingual diagram, chart, and technical drawing comprehension
- Significantly higher accuracy on complex visual question answering across 23 languages
- Open weights available for private corporate hosting
Limitations & Trade-offs
- 16K context limit requires chunking for massive multi-page documents
- Video streams not supported
Editorial Assessment
The most capable open multilingual vision-language model in the 30B class.
Best for: International patent document inspection, global medical chart OCR, and multilingual enterprise analytics
Compare Aya Vision 32B
Frequently Asked Questions about Aya Vision 32B
How much does the Aya Vision 32B API cost per 1M tokens?+
Aya Vision 32B by Cohere is priced at $0.35 per 1M input tokens and $0.8 per 1M output tokens. With prompt caching enabled, cached input tokens cost $0.08 per 1M tokens, and asynchronous Batch API requests receive a 50% discount.
Should I use a $20/month subscription or the Aya Vision 32B pay-as-you-go API?+
A flat $20/month consumer subscription is predictable for daily interactive chat and manual coding. However, if you are building software, running automated workflows, or sending low-to-moderate volume with prompt caching enabled, the pay-as-you-go API is typically far more cost-effective and offers higher rate limits without manual UI caps.
How many words and pages fit in Aya Vision 32B's context window?+
Aya Vision 32B features a 16K tokens context window, which equals approximately 12,000 English words or ~32 standard single-spaced pages in a single prompt. Its maximum output generation per turn is 4,096 tokens.
Can I run Aya Vision 32B locally on my own GPU (Open Weights)?+
Yes, Aya Vision 32B (32B) releases open weights that can be self-hosted locally or on private cloud GPUs using vLLM, Ollama, SGLang, or llama.cpp.
Does Cohere train on data sent to the Aya Vision 32B API, and is commercial use allowed?+
100% Isolated when self-hosted on private GPUs or VPC. License tier: Open-Weights Research & Commercial License. Full commercial ownership of generated outputs. Supported deployment channels include: Self-Hosted GPU (vLLM / Ollama), Private VPC, or Managed Serverless API.
How does Aya Vision 32B perform on coding and reasoning benchmarks?+
Aya Vision 32B achieves a composite Intelligence Index of 45/100, scoring 32% on SWE-bench Verified, 51.5% on GPQA Diamond, 74% on MMLU-Pro, and 1280 pts on Chatbot Arena Elo.
What is Aya Vision 32B best used for?+
Aya Vision 32B is best suited for: International patent document inspection, global medical chart OCR, and multilingual enterprise analytics. Assessment: The most capable open multilingual vision-language model in the 30B class.
Cohere Family Lineage, Routing Tiers & Upgrade Path
Step down to a lighter sibling for high-volume routing or step up to Cohere's flagship reasoning tier:
North Mini Code
Command A Plus
Command A Reasoning
Command A Translate
Command A Vision
Command A
AI Glossary & Architecture Concepts
View full glossary →Key machine learning concepts and benchmarks relevant to Aya Vision 32B:
The basic unit of text that a language model processes, typically representing about 3 to 4 English characters or roughly 0.75 words. For example, the sentence "Artificial intelligence is fast" breaks down into about 5 tokens.
A standardized evaluation suite used to objectively measure and rank an AI model's capabilities across specific tasks — such as SWE-bench Verified for software engineering, AIME for competitive math, or MMLU for general knowledge.
The maximum number of tokens a model can consider at once — both the input it reads and the output it writes. A 1,000,000-token context window can hold roughly 750,000 English words in a single request.
The ability of a language model to recognize when an external capability is needed and output structured JSON arguments to invoke external APIs, databases, or terminal commands. The runtime executes the tool and passes results back to the model.
An open standard created by Anthropic that allows AI applications and agents to connect seamlessly to local filesystems, development environments, and third-party APIs via standardized server interfaces.
A cost and latency optimization where inference providers store pre-computed attention states for repeated prompt prefixes. Reusing cached tokens typically slashes API input costs by 50% to 90% and cuts TTFT substantially.
Benchmark scores, token prices, and rate limits evolve rapidly as AI labs release model updates. Always verify current specifications and pricing on the official provider documentation.