Models
Every model we serve, with the numbers.
Claude and GPT models by their normal ids. Standard-tier models cost half the vendor list price; priority-tier variants are billed at a separate rate. Prices are USD per 1M tokens, excl. VAT, identical for API and chat.
Anthropic
9 standard-tier Anthropic models at half the list price.
Claude Fable 5.1NewPriority
- Vendor
- Anthropic
- Context
- 200K
- Max output
- 32K
- Streaming
- Tool calls
- Vision
- Reasoning
- Structured outputs
- Prompt caching
Claude Fable 5
- Vendor
- Anthropic
- Context
- 200K
- Max output
- 32K
- Streaming
- Tool calls
- Vision
- Reasoning
- Structured outputs
- Prompt caching
Claude Opus 5
- Vendor
- Anthropic
- Context
- 200K
- Max output
- 32K
- Streaming
- Tool calls
- Vision
- Reasoning
- Structured outputs
- Prompt caching
Claude Sonnet 5
- Vendor
- Anthropic
- Context
- 200K
- Max output
- 32K
- Streaming
- Tool calls
- Vision
- Structured outputs
- Prompt caching
Claude Opus 4.8
- Vendor
- Anthropic
- Context
- 200K
- Max output
- 32K
- Streaming
- Tool calls
- Vision
- Reasoning
- Structured outputs
- Prompt caching
Claude Opus 4.7
- Vendor
- Anthropic
- Context
- 200K
- Max output
- 32K
- Streaming
- Tool calls
- Vision
- Reasoning
- Structured outputs
- Prompt caching
Claude Opus 4.6
- Vendor
- Anthropic
- Context
- 200K
- Max output
- 32K
- Streaming
- Tool calls
- Vision
- Reasoning
- Structured outputs
- Prompt caching
Claude Sonnet 4.6
- Vendor
- Anthropic
- Context
- 200K
- Max output
- 32K
- Streaming
- Tool calls
- Vision
- Structured outputs
- Prompt caching
Claude Haiku 4.5
- Vendor
- Anthropic
- Context
- 200K
- Max output
- 32K
- Streaming
- Tool calls
- Vision
- Structured outputs
- Prompt caching
OpenAI
4 standard-tier OpenAI models at half the list price.
GPT-6 AstraNew
- Vendor
- OpenAI
- Context
- 400K
- Max output
- 128K
- Streaming
- Tool calls
- Reasoning
- Structured outputs
- Prompt caching
GPT-5.6 Sol
- Vendor
- OpenAI
- Context
- 400K
- Max output
- 128K
- Streaming
- Tool calls
- Reasoning
- Structured outputs
- Prompt caching
GPT-5.6 Terra
- Vendor
- OpenAI
- Context
- 400K
- Max output
- 128K
- Streaming
- Tool calls
- Reasoning
- Structured outputs
- Prompt caching
GPT-5.5
- Vendor
- OpenAI
- Context
- 400K
- Max output
- 128K
- Streaming
- Tool calls
- Reasoning
- Structured outputs
- Prompt caching
Priority
Priority routing and higher cache efficiency for coding agents; billed at the priority rate.
Claude Fable 5 PriorityPriority
- Vendor
- Anthropic
- Context
- 200K
- Max output
- 32K
- Streaming
- Tool calls
- Vision
- Reasoning
- Structured outputs
- Prompt caching
Claude Opus 5 PriorityPriority
- Vendor
- Anthropic
- Context
- 200K
- Max output
- 32K
- Streaming
- Tool calls
- Vision
- Reasoning
- Structured outputs
- Prompt caching
Claude Sonnet 5 PriorityPriority
- Vendor
- Anthropic
- Context
- 200K
- Max output
- 32K
- Streaming
- Tool calls
- Vision
- Structured outputs
- Prompt caching
Claude Opus 4.8 PriorityPriority
- Vendor
- Anthropic
- Context
- 200K
- Max output
- 32K
- Streaming
- Tool calls
- Vision
- Reasoning
- Structured outputs
- Prompt caching
Claude Opus 4.7 PriorityPriority
- Vendor
- Anthropic
- Context
- 200K
- Max output
- 32K
- Streaming
- Tool calls
- Vision
- Reasoning
- Structured outputs
- Prompt caching
Claude Opus 4.6 PriorityPriority
- Vendor
- Anthropic
- Context
- 200K
- Max output
- 32K
- Streaming
- Tool calls
- Vision
- Reasoning
- Structured outputs
- Prompt caching
Claude Sonnet 4.6 PriorityPriority
- Vendor
- Anthropic
- Context
- 200K
- Max output
- 32K
- Streaming
- Tool calls
- Vision
- Structured outputs
- Prompt caching
Claude Haiku 4.5 PriorityPriority
- Vendor
- Anthropic
- Context
- 200K
- Max output
- 32K
- Streaming
- Tool calls
- Vision
- Structured outputs
- Prompt caching
Claude Haiku 4.5 20251001Priority
- Vendor
- Anthropic
- Context
- 200K
- Max output
- 32K
- Streaming
- Tool calls
- Vision
- Structured outputs
- Prompt caching
Endpoints and compatibility
All models are reachable through the OpenAI-compatible /v1/chat/completions and /v1/responses endpoints and through the Anthropic-compatible /v1/messages endpoint at https://api.halfbill.uk. Each model page shows both request shapes for that model id.
Halfbill is an independent service and is not affiliated with, endorsed by, or sponsored by OpenAI or Anthropic. Model names are trademarks of their respective owners.