LLM compare
LLM compare
Claude, GPT, Gemini, Mistral, Llama — price, context window, vision, EU hosting.
| Model | $ / 1M in | $ / 1M out | Ctx | Vision | EU host |
|---|---|---|---|---|---|
Claude Fable 5 Anthropic Most capable for long-horizon reasoning. Requires 30-day retention. | $10 | $50 | 1,000k | ✓ | EU |
Claude Opus 5 Anthropic Best for reasoning and coding. EU hosting available. | $5 | $25 | 1,000k | ✓ | EU |
Claude Sonnet 5 Anthropic Best price/performance for production. | $2 | $10 | 1,000k | ✓ | EU |
GPT-5.6 Sol OpenAI Excellent at function calling. Priciest output here. | $5 | $30 | 1,050k | ✓ | — |
GPT-5.6 Luna OpenAI Very cheap at volume. Weaker on reasoning. | $0.2 | $1.2 | 1,050k | ✓ | — |
Gemini 3.1 Pro Long context, tight Google Cloud fit. Surcharge above 200k. | $2 | $12 | 1,000k | ✓ | EU |
Mistral Large 3 Mistral EU-native and the cheapest here. Ideal for GDPR-sensitive work. | $0.5 | $1.5 | 256k | — | EU |
Llama 4 Maverick Meta (self-hosted) Open weights, self-hosted in the EU if you have GPU infra. Cost = electricity + ops. | $0 | $0 | 1,000k | ✓ | EU |