Foundation Models

Foundation models are large AI models trained on broad data that can be adapted to many downstream tasks. The term encompasses LLMs, vision models, and multimodal systems.

Commercial Models

Anthropic Claude

Model	Context	Strengths
Claude 4.5 Sonnet	200K (1M beta)	Best for coding, agentic tasks
Claude 4.5 Opus	200K	Premium intelligence, deep reasoning
Claude 4.5 Haiku	200K	Fastest, near-frontier intelligence

• Extended thinking for complex reasoning
• 64K max output tokens
• Strong instruction following and tool use
• Constitutional AI training
• API Documentation

OpenAI GPT

Model	Context	Strengths
GPT-5.2	128K+	Best for coding, agentic tasks
GPT-5 mini	128K	Faster, cost-efficient
GPT-5 nano	128K	Fastest, most cost-efficient
GPT-4.1	128K	Smartest non-reasoning model
o3 / o4-mini	128K	Deep reasoning, STEM

• Open-weight models available (gpt-oss-120b, gpt-oss-20b)
• Sora 2 for video generation
• Strong function calling and tool use
• API Documentation

Google Gemini

Model	Context	Strengths
Gemini 3 Pro	1M+	Most intelligent, complex tasks
Gemini 3 Flash	1M	Frontier intelligence at speed
Gemini 2.5 Flash-Lite	1M	High volume, cost efficient

• State-of-the-art reasoning and multimodal
• Extended thinking capabilities
• Strong agentic and coding performance
• API Documentation

Others

• Cohere Command — Enterprise focus, RAG-optimised
• Amazon Nova — AWS Bedrock integration
• xAI Grok — Strong reasoning, real-time data

Open-Source Models

Meta Llama

Model	Parameters	Context	Notes
Llama 4 Scout	17B (16 experts)	128K	Multimodal MoE
Llama 4 Maverick	17B (128 experts)	128K	Larger expert pool
Llama 3.3 70B	70B	128K	Text-only instruct

• Mixture-of-experts architecture
• Native multimodal (text + images)
• Permissive license (with restrictions)
• Llama Downloads

Mistral

Model	Parameters	Context	Notes
Mistral Large	~100B	128K	Commercial flagship
Mixtral 8x22B	MoE	64K	Mixture of experts
Mistral OCR 3	—	—	Document processing

• European AI company
• Strong efficiency/performance ratio
• Le Chat consumer product
• Mistral AI

Qwen (Alibaba)

Model	Parameters	Context	Notes
Qwen3	Various	128K	Strong all-round
Qwen3-VL	Various	128K	Vision-language
Qwen3-TTS	Various	—	Text-to-speech

• Strong multilingual (especially Chinese)
• Extensive model family (400+ variants)
• Embedding, reranking, and omni models
• Qwen

DeepSeek

Model	Parameters	Notes
DeepSeek-V3.2	685B MoE	Latest flagship
DeepSeek R1	Various	Strong reasoning
DeepSeek Coder	Various	Code-specialised

• Competitive with frontier models
• Cost-efficient training and inference
• Open weights available
• DeepSeek

Others

• Yi (01.AI) — Strong multilingual
• Phi (Microsoft) — Small but capable
• Gemma 3 (Google) — Open weights, research-friendly
• OLMo (AI2) — Fully open including training data
• Grok (xAI) — Available via API

Model Comparison Factors

Capability Benchmarks

See Evaluation & Benchmarking for details.

• MMLU — Broad knowledge
• HumanEval — Coding
• GSM8K — Math reasoning
• GPQA — Graduate-level science

Practical Considerations

Factor	Considerations
Latency	Time to first token, tokens/second
Cost	Per-token pricing, volume discounts
Context	How much text can be processed
Reliability	Uptime, consistency
Privacy	Data handling, compliance
Ecosystem	SDKs, documentation, support

License Types

• Proprietary API — No access to weights (GPT-5, Claude)
• Gated open — Weights available with restrictions (Llama 4)
• Permissive open — Few restrictions (Mistral, Qwen, DeepSeek)
• Fully open — Weights, code, and training data (OLMo)

API Providers

Model Providers

Direct from the source:

• OpenAI API
• Anthropic API
• Google AI Studio
• Mistral API
• Cohere

Aggregators / Routers

Access multiple models through one API:

• OpenRouter — Multiple providers, unified API
• Together AI — Open models, fine-tuning
• Fireworks AI — Fast inference
• Groq — Ultra-fast inference
• Anyscale

Cloud Platforms

• AWS Bedrock — Multiple providers in AWS
• Azure OpenAI — OpenAI in Azure
• Google Vertex AI — Gemini and others

Choosing a Model

Decision Framework

1. Task requirements — What capability is most important?
2. Latency needs — Real-time vs batch processing
3. Cost constraints — Budget per million tokens
4. Privacy requirements — Can data leave your environment?
5. Context needs — How much text per request?
6. Compliance — Regulatory requirements

Rules of Thumb

• Start with a capable model (Claude 4.5 Sonnet, GPT-5.2, Gemini 3 Pro)
• Optimise for cost/speed once it works (mini/nano/Flash variants)
• Open models for privacy-sensitive use cases (Llama 4, Qwen3, DeepSeek)
• Smaller models for high-volume, simple tasks

Staying Current

The landscape changes rapidly. Track developments:

• Artificial Analysis — Performance benchmarks
• LMSYS Chatbot Arena — Human preference rankings
• Hugging Face Open LLM Leaderboard
• Model provider blogs and announcements

Resources

• Foundation Models Paper — Stanford’s influential survey
• State of GPT — Karpathy’s overview
• Open LLM Leaderboard
• Chatbot Arena Leaderboard