Anthropic · checked October 13, 2026
How do I run Claude Opus 5.5 in Claude Code, OpenCode or Codex?
Claude Opus 5.5 runs on Standard Compute at the base URL https://api.stdcmpt.com/v1 with the exact model ID anthropic/claude-opus-5.5. The same key works in Claude Code, OpenCode and Codex.
| Model | Anthropic's Opus model in the Claude 5.5 generation, which Anthropic positions for long-running agentic coding and knowledge work. It has a 1M-token context window, 128K max output and adaptive thinking that is always on. |
|---|---|
| Model ID | anthropic/claude-opus-5.5Short form claude-opus-5.5 also works |
| Base URL | https://api.stdcmpt.com/v1Claude Code uses https://api.stdcmpt.com |
| Context | 1M tokens; about 850K input and 50K output per request |
| Input | Text, images and files |
| Reasoning | low, medium, high, xhigh or max, set per model on the Models page |
Copy a working config
Pick your agent. Each config pins Opus 5.5 for the main conversation; leave the model as standardcompute instead if you want routing to decide when Opus is worth it.
# Install the Standard Compute launcher once
curl -fsSL https://standardcompute.com/cli/install.sh | sh
# Start Claude Code on your key, pinned to anthropic/claude-opus-5.5
standardcompute launch claude -- --model anthropic/claude-opus-5.5
# Or configure it yourself. No /v1 here: Claude Code adds /v1/messages.
export ANTHROPIC_BASE_URL=https://api.stdcmpt.com
export ANTHROPIC_AUTH_TOKEN=YOUR_API_KEY
export ANTHROPIC_MODEL=anthropic/claude-opus-5.5
claude
Replace YOUR_API_KEY with a key from the dashboard; OpenCode reads it from STANDARDCOMPUTE_API_KEY. Full setup: Claude Code, OpenCode, Codex CLI.
What it costs
| Per 1M tokens | Price |
|---|---|
| Input | $4.00 |
| Cache hit | $0.20 |
| 5-minute cache write | $5.00 |
| Output | $20.00 |
Anthropic standard API rates. The full 1M window bills at the same rate. Fast mode costs $8 / $40 and is a first-party Anthropic option. Source: Anthropic API pricing, checked October 13, 2026.
Averages from coding-agent traffic on our gateway, Sep 13 to Oct 12, 2026: 87% of prompt tokens are cache hits and a turn returns about 700 tokens. Vendor list prices.
The $19 Starter plan includes a $20 compute budget, roughly 260 turns like this one. That is enough to put Opus on hard problems; for a working month of Opus-only sessions, a larger plan or a mixed selection fits better.
What works differently here
- Claude Code's thinking toggle does nothing here. The Messages endpoint ignores the thinking parameter and returns no thinking blocks, so how hard Opus 5.5 reasons comes from the default you set under Customize on the Models page.
- With the launcher, only the main conversation is pinned. Claude Code's subagents and quick background calls still ask for its built-in sonnet and haiku names, and Standard Compute treats those as smart routing across the models you selected.
- A Standard Compute plan is not a Claude Pro or Max subscription, and Anthropic's Fast mode is not offered. Requests also stop at about 850K input tokens, a little short of the full 1M window.
Other ways to run Claude Opus 5.5
| Anthropic API $4 / $20 per 1M | First-party, pay per token. Batch requests are 50% off and prompt caching is yours to set up. |
|---|---|
| Claude Pro / Max Subscription | Claude Code's native login. Usage runs in session and weekly limits rather than a token bill. |
| Bedrock / Vertex / Foundry Pay per token | Cloud billing for procurement teams. On Bedrock the ID is anthropic.claude-opus-5-5. |
| Standard Compute Monthly plan | One key for Claude Code, Codex and OpenCode, drawing on a monthly compute budget. |
FAQ
Why does claude-opus-5-5 return model_not_found?
That is Anthropic's own ID, with hyphens. Standard Compute uses the dotted form, anthropic/claude-opus-5.5, so model IDs match across every vendor in the catalog.
Do I need a Claude subscription to use Opus 5.5 in Claude Code?
No. Claude Code only needs a base URL and a token that speak the Messages API. Your Claude login stays as it is; running claude without the launcher still uses it.
Is Opus 5.5 cheaper to run than Opus 5?
Yes, on Anthropic's list prices: $4 / $20 per million tokens against $5 / $25, and cache hits cost 5% of the input price on Opus 5.5 against 10% on Opus 5. On an agent with a mostly cached prompt, the cache rate matters as much as the headline price.
Should I pin Opus 5.5 or let routing choose?
Pin it for a session where every step is hard. Otherwise select Opus 5.5 next to an efficient model on the Models page and keep the model set to standardcompute: routing sends hard requests to Opus and routine ones to the cheaper model, so the budget lasts longer.
Plans from $19/mo
One key for Claude Opus 5.5 and the rest of the catalog, with a monthly compute budget you can watch in the dashboard and no overage charges.
Sources: Anthropic API pricing · Anthropic models overview
Related: Claude Opus 5 setup · Cheapest way to run Claude Code · All model IDs