Migrating from Anthropic

ORYVOX exposes Claude models through the OpenAI-compatible API, so you switch from the Messages API to chat completions (or /responses).

Request mapping

Anthropic Messages	ORYVOX `/chat/completions`
`model: "claude-sonnet-5-5"`	`model: "anthropic/claude-sonnet-5.5"` (or a routing profile)
`system: "..."`	first message `{"role": "system", "content": "..."}`
`messages[].content` blocks (`text`, `image`)	string, or parts `{"type":"text"}` / `{"type":"image_url"}`
`max_tokens` (required)	`max_tokens` (optional)
`tools[].input_schema`	`tools[].function.parameters`
`tool_use` block	`message.tool_calls[]` (`function.arguments` is a JSON string)
`tool_result` block	message `{"role": "tool", "tool_call_id": ..., "content": ...}`
`stop_reason: "end_turn" / "tool_use" / "max_tokens"`	`finish_reason: "stop" / "tool_calls" / "length"`
`usage.input_tokens / output_tokens`	`usage.prompt_tokens / completion_tokens`
`x-api-key` header	`Authorization: Bearer ory_live_...`

Before and after

# Before
import anthropic
client = anthropic.Anthropic()
msg = client.messages.create(
    model="claude-sonnet-5-5", max_tokens=500, system="Be concise.",
    messages=[{"role": "user", "content": "Summarise this ticket: ..."}],
)
print(msg.content[0].text)

# After
from oryvox import Oryvox
client = Oryvox()
res = client.responses.create(
    model="anthropic/claude-sonnet-5.5", instructions="Be concise.",
    input="Summarise this ticket: ...", max_output_tokens=500,
)
print(res.output_text)

Streaming

Anthropic's content_block_delta events become response.output_text.delta events (/responses) or chat.completion.chunk deltas (/chat/completions). See Streaming.

Notes

Anthropic prompt caching is applied by ORYVOX automatically where it helps; you do not need cache_control blocks. Discounts appear in cost.
Extended thinking: use oryvox-reasoning or a reasoning-capable Claude model.
With BYOK you can keep using your Anthropic account and pricing.