- Context
- 10,000,000
- Max output
- —
- Family
- Llama 4
- Input
- Text, Image
Reference view
Model Context Windows
Compare the documented context and output limits of indexed models, with provider, family, modality and API context from the iilib knowledge graph.
A larger context window is a capacity limit, not a quality score. Effective usable context, pricing, latency, tool behavior and provider-specific restrictions can differ by workload and API.
Specification table
Context and output limits
Rows are sorted by documented context-window size. Missing values remain visible as unknown rather than inferred.
- Context
- 1,050,000
- Max output
- 128,000
- Family
- GPT-5.6
- Input
- Text, Image
- Context
- 1,050,000
- Max output
- 128,000
- Family
- GPT-5.6
- Input
- Text, Image
- Context
- 1,050,000
- Max output
- 128,000
- Family
- GPT-5.6
- Input
- Text, Image
- Context
- 1,050,000
- Max output
- 128,000
- Family
- GPT-6
- Input
- Text, Image
- Context
- 1,050,000
- Max output
- 128,000
- Family
- GPT-6
- Input
- Text, Image
- Context
- 1,050,000
- Max output
- 128,000
- Family
- GPT-6
- Input
- Text, Image
- Context
- 1,050,000
- Max output
- 128,000
- Family
- GPT-6
- Input
- Text, Image
- Context
- 1,050,000
- Max output
- 128,000
- Family
- GPT-6
- Input
- Text, Image
- Context
- 1,050,000
- Max output
- 128,000
- Family
- GPT-6
- Input
- Text, Image
- Context
- 1,050,000
- Max output
- 128,000
- Family
- GPT-6
- Input
- Text, Image
- Context
- 1,048,576
- Max output
- 65,536
- Family
- Gemini 3
- Input
- Text, Image, Audio, Video, File
- Context
- 1,048,576
- Max output
- 65,536
- Family
- Gemini 4
- Input
- Text, Image, Audio, Video, File
- Context
- 1,048,576
- Max output
- —
- Family
- GLM-5.3
- Input
- Text
- Context
- 1,048,576
- Max output
- —
- Family
- GLM-5.3
- Input
- Text, Image, Video
- Context
- 1,048,576
- Max output
- —
- Family
- Kimi K3
- Input
- Text, Image, Video
- Context
- 1,000,000
- Max output
- 128,000
- Family
- Claude 5.1
- Input
- Text, Image
- Context
- 1,000,000
- Max output
- 128,000
- Family
- Claude 5.2
- Input
- Text, Image
- Context
- 1,000,000
- Max output
- 128,000
- Family
- Claude 5.5
- Input
- Text, Image
- Context
- 1,000,000
- Max output
- 128,000
- Family
- Claude 5.5
- Input
- Text, Image
- Context
- 1,000,000
- Max output
- 384,000
- Family
- DeepSeek V4
- Input
- Text
- Context
- 1,000,000
- Max output
- 384,000
- Family
- DeepSeek V4
- Input
- Text, Image
- Context
- 1,000,000
- Max output
- —
- Family
- Llama 4
- Input
- Text, Image
- Context
- 1,000,000
- Max output
- —
- Family
- Mistral Large
- Input
- Text, Image
- Context
- 1,000,000
- Max output
- —
- Family
- Nemotron 3.5
- Input
- Text
- Context
- 1,000,000
- Max output
- 65,536
- Family
- Qwen3-Coder
- Input
- Text
- Context
- 500,000
- Max output
- —
- Family
- Grok 4
- Input
- Text, Image
- Context
- 256,000
- Max output
- —
- Family
- Gemma 4
- Input
- Text, Image, Audio
- Context
- 256,000
- Max output
- —
- Family
- Gemma 4
- Input
- Text, Image
- Context
- 256,000
- Max output
- —
- Family
- Mistral Medium
- Input
- Text, Image
- Context
- 131,072
- Max output
- —
- Family
- gpt-oss
- Input
- Text
- Context
- 128,000
- Max output
- —
- Family
- DeepSeek R1
- Input
- Text
- Context
- 128,000
- Max output
- —
- Family
- DeepSeek V3
- Input
- Text
- Context
- 128,000
- Max output
- —
- Family
- Qwen3
- Input
- Text
- Context
- 128,000
- Max output
- —
- Family
- Qwen3
- Input
- Text
- Context
- 65,536
- Max output
- —
- Family
- Olmo 3
- Input
- Text
- Context
- 65,536
- Max output
- —
- Family
- Olmo 3
- Input
- Text
- Context
- 32,000
- Max output
- —
- Family
- Qwen3
- Input
- Text
- Context
- 32,000
- Max output
- —
- Family
- Qwen3 Embedding
- Input
- Text
- Context
- 32,000
- Max output
- —
- Family
- Qwen3 Reranker
- Input
- Text
- Context
- —
- Max output
- —
- Family
- GPT Image 2.6
- Input
- Text, Image
- Context
- —
- Max output
- —
- Family
- GPT Image 2.6
- Input
- Text, Image
Coverage
19 models with output-limit data
Unknown values are not filled from estimates. The table changes only when the underlying model record is updated from a primary source.
Cost context
Context size is only one constraint
For token pricing alongside these limits, use the graph-derived pricing reference.
Open LLM API Pricing →