Claude Opus 4.5 (2025-11-01)
How TheRouter serves this differently from the vendor
Anthropic's own API exposes this model under the id 'claude-opus-4-5-20251101' (hyphens, no dots). Callers must use the Anthropic SDK or native API format, supply an Anthropic API key, and pass the effort parameter directly via the 'thinking' object or the effort field. The model is identical in capability to the rolling 'claude-opus-4.5' alias at the November 2025 weights.
TheRouter proxies this model through the standard OpenAI-compatible /v1/chat/completions endpoint under the id 'anthropic/claude-opus-4.5-20251101' (dots, slash-separated provider prefix). Callers use any OpenAI SDK, supply a single TheRouter API key, and pass reasoning_effort instead of the native effort field. TheRouter translates the request to Anthropic's API format and returns an OpenAI-shaped response. No Anthropic SDK or direct Anthropic API key is required.
API guide
Chat (pinned)
Pin to the November 2025 snapshot for reproducible, auditable results.
curl https://api.therouter.ai/v1/chat/completions \
-H "Authorization: Bearer $THEROUTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "anthropic/claude-opus-4.5-20251101",
"messages": [{"role":"user","content":"Refactor this repo to use the new auth middleware."}],
"reasoning_effort": "high"
}'Tools / Agents
Use the pinned snapshot id for evaluation harnesses and A/B tests where the agent loop must stay on a fixed model version.
curl https://api.therouter.ai/v1/chat/completions \
-H "Authorization: Bearer $THEROUTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "anthropic/claude-opus-4.5-20251101",
"messages": [{"role":"user","content":"List all test files in this repo."}],
"tools": [{"type":"function","function":{"name":"read_file","description":"Read a file","parameters":{"type":"object","properties":{"path":{"type":"string"}},"required":["path"]}}}],
"reasoning_effort": "high"
}'Fact ledger — every claim on this page traces here
| source | URL | retrieved | |
|---|---|---|---|
| Pinned release date | www.anthropic.com ↗ | 2026-08-20 | verified |
| Anthropic model id | www.anthropic.com ↗ | 2026-08-20 | verified |
| Canonical (rolling) alias | docs.anthropic.com ↗ | 2026-08-20 | verified |
| Training cutoff | — | — | unknown |
| Reasoning controls | platform.claude.com ↗ | 2026-08-20 | verified |
| Deployment | aws.amazon.com ↗ | 2026-08-20 | verified |
| License | — | — | verified |
| SWE-bench Verified | www.anthropic.com ↗ | 2026-08-20 | verified |
| GPQA Diamond | anthropic.com ↗ | 2026-08-20 | verified |
| ARC-AGI-2 | anthropic.com ↗ | 2026-08-20 | verified |
| Prompt injection attack success (Gray Swan) | www.anthropic.com ↗ | 2026-08-20 | verified |
| Claude Opus 4.5 released — first model above 80 % on SWE-bench Verified | anthropic.com/news ↗ | 2026-08-20 | verified |
| Claude Opus 4.5 available in Amazon Bedrock with global cross-region inference | aws.amazon.com ↗ | 2026-08-20 | verified |
| What is the difference between anthropic/claude-opus-4.5 and anthropic/claude-opus-4.5-20251101? | docs.anthropic.com ↗ | 2026-08-20 | to verify |
| When will claude-opus-4-5-20251101 be deprecated? | docs.anthropic.com ↗ | 2026-08-20 | to verify |
| Does claude-opus-4.5-20251101 support the effort parameter and manual thinking budget_tokens? | platform.claude.com ↗ | 2026-08-20 | to verify |