Claude Mythos 5.1

Claude Mythos 5.1 is Anthropic's model for long-running agentic coding and knowledge work. Send the official Anthropic Messages request to SeedRouter: change the base URL and the API key, keep the body. The same model also answers the OpenAI Chat Completions and Responses formats.

Model ID

See the model page for current prices.

Quick example

curl https://api.seedrouter.ai/v1/messages \
  -H "x-api-key: $SEEDROUTER_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-mythos-5-1",
    "max_tokens": 1024,
    "messages": [{"role": "user", "content": "Summarize the trade-offs of event sourcing in three bullet points."}]
  }'

Endpoints

The Messages endpoint forwards your request body as sent, including optional fields, and returns the official response. An anthropic-beta header is passed on as well. Keep the API key in server-side code.

Parameters

The current native contract has 24 top-level fields. Unknown fields and unsupported sampling settings (temperature, top_p, top_k) are ignored before forwarding. Invalid values of supported fields are rejected.

Thinking, tools and state

Claude Mythos 5.1 uses adaptive thinking with high effort by default. It does not support enabled, disabled or between_tools thinking, or budget_tokens. thinking.display: updates needs thinking-display-updates-2026-08-18. thinking.block_binding allows prefix_mismatch_behavior error, drop_block or null with thinking-binding-controls-2026-08-01.

output_config.task_budget requires task-budgets-2026-03-13, type tokens, integer total of at least 20000, and optional nonnegative integer remaining. Omit remaining when using compaction or replaying a signed compaction block. On-demand compaction requires compact-2026-09-04 and cannot accompany context_management, stop_sequences, structured output or forced tools. Threshold compaction uses compact-2026-01-12, with an input_tokens trigger of at least 50000.

Per-message effort on system messages requires mid-conversation-output-config-2026-07-01. Final assistant prefill is rejected; official pause_turn and compaction continuations remain valid.

Computer use accepts computer_toolset_20260801 or computer_20251124 with computer-use-2025-11-24. Advisor tools require advisor-tool-2026-03-01 and a supported target. Your application controls tool execution and permissions.

Thread continuation sends only new messages and must omit system and tools. It cannot accompany compaction or context edits other than keeping all thinking. State expires after five minutes without a request; retain your conversation locally. State-dependent constraints are checked when continuing.

Mythos fallback targets depend on official model and account capabilities; a target list has not been established by the local verification. Advanced beta execution and fast-mode charging require live verification; the request schema is not a measured capability guarantee.

Media inputs

Images and PDFs go in a user turn's content as image and document blocks, with a url source or, as in the official API, a base64 source:

{"role": "user", "content": [
  {"type": "image", "source": {"type": "url", "url": "https://example.com/chart.png"}},
  {"type": "text", "text": "What does this chart show?"}
]}

Replace the example URL with a publicly reachable file of your own.

Billing dimensions

See the current rates on the model page. A request is billed by the tokens it uses:

• input tokens,
• output tokens, including thinking,
• prompt-cache reads, and
• prompt-cache writes, with separate 5-minute and 1-hour rates.

The charge is taken from the usage reported with the finished response. A request that fails is not charged. Your account's usage records show the exact charge for every request.

Output

A non-streaming request returns the official message object:

{
  "id": "msg_...",
  "type": "message",
  "role": "assistant",
  "model": "claude-mythos-5-1",
  "content": [{"type": "text", "text": "..."}],
  "stop_reason": "end_turn",
  "usage": {"input_tokens": 18, "output_tokens": 4, "cache_read_input_tokens": 0, "cache_creation_input_tokens": 0}
}

With "stream": true the response is a stream of the official events: message_start, content_block_start, content_block_delta, content_block_stop, message_delta and message_stop. The final message_delta carries the output token count.

OpenAI-compatible formats

The same model answers the OpenAI formats, so existing OpenAI code only needs a new base URL and model ID:

from openai import OpenAI

client = OpenAI(api_key="YOUR_SEEDROUTER_KEY", base_url="https://api.seedrouter.ai/v1")

chat = client.chat.completions.create(
    model="claude-mythos-5-1",
    messages=[{"role": "user", "content": "Hello"}],
)
response = client.responses.create(model="claude-mythos-5-1", input="Hello")

These requests are converted to the Messages format, so a field with no Messages counterpart has no effect. Their responses carry the official OpenAI fields and may include a few extra usage fields; read usage.total_tokens and the official fields.

Errors

Errors on /v1/messages use the Anthropic shape, {"type": "error", "error": {"type": "...", "message": "..."}}; the other formats use {"error": {"code": ..., "message": "..."}}. The code is a code from the common error catalog. Failed requests are not charged.

Tips

• Start with the default effort and raise it only for tasks that need more thinking; effort changes both quality and cost.
• Set max_tokens high enough for the thinking as well as the answer: it is one budget for both.
• Put long, reused context first and mark it with cache_control so later requests read it from the cache at the lower rate.

Related

• Claude Mythos 5.1 playground and pricing
• Download OpenAPI
• Copyable Markdown

Model tiers

Use claude-mythos-5-1 for Default. Additional tiers appear automatically in the live price table when configured. Send the complete model:group ID shown on a tier to select it. Your key and account must permit that tier; its prices apply. The playground loads available choices and estimates without requiring you to paste a key.

The optional X-SeedRouter-Group header remains supported. If supplied together with a model suffix, both must select the same group.

Check access without generating a response:

curl "https://api.seedrouter.ai/v1/models" \
  -H "Authorization: Bearer $SEEDROUTER_API_KEY"
curl "https://api.seedrouter.ai/api/seedrouter/llm/groups?model=claude-mythos-5-1" \
  -H "Authorization: Bearer $SEEDROUTER_API_KEY"

Official references

• Model overview
• Messages reference
• Thinking

Contract checked November 10, 2026. Current prices are on the model page.