Claude Opus 5.5 on QuickSilver Pro
Claude Opus 5.5 is the first model of Anthropic's Claude 5.5 family and the most capable Claude to date, ahead of Claude Opus 5 and Claude Fable 5.1 on agentic coding, knowledge work and computer use. It takes text and image input across a 1M-token context with up to 128K output tokens. QuickSilver Pro charges $3.20 input / $16.00 output per million tokens, 20% below Anthropic's $4 / $20 list price.
At a glance
Anthropic's most capable Claude for agentic coding, knowledge work and computer use, with a 1M-token context.
Pricing comparison ($/1M tokens)
| Provider | Input | Output | vs QSP |
|---|---|---|---|
| QuickSilver Pro | $3.20 | $16.00 | lowest-cost |
| Anthropic list price (Anthropic API) | $4.00 | $20.00 | 20% lower |
When to use
Reach for Claude Opus 5.5 when the task is the hard part: long-running coding agents working in a terminal over a large repository, multi-step knowledge work that ends in a finished deliverable, and computer-use agents that operate software through screenshots. It supports streaming, tool calling, JSON Schema output and image input on both Chat Completions and Responses, and it is 20% cheaper per token than Claude Opus 5.
When to use something else
For routine chat, extraction or high-volume automation, a smaller model such as Claude Sonnet 5 or Claude Haiku 4.5 costs a fraction as much. The model reasons before it answers, and reasoning tokens bill as output, so leave generous room in `max_tokens` on hard problems. `temperature` and `top_p` are not applied on this model; requests that send them still succeed.
Quickstart (curl)
curl https://api.quicksilverpro.io/v1/chat/completions \
-H "Authorization: Bearer $QSP_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "claude-opus-5-5",
"messages": [{"role": "user", "content": "Hello!"}]
}'OpenAI-compatible. One-line migration via base_url.
FAQ
$3.20 per million input tokens and $16.00 per million output tokens, 20% below Anthropic's $4 / $20 list price. Reasoning tokens bill as output tokens, and usage.cost in each response tells you what the call charged.
On published launch benchmarks it leads both. Terminal-Bench 4.0, which measures agentic coding: 66.4% for Opus 5.5, 55.8% for Fable 5.1 and 52.3% for Opus 5. It also scores 1846 on GDPval-AA for knowledge work and 81.8% on OSWorld 2.0 for computer use. It ranks #1 on the Artificial Analysis Intelligence Index v4.3 with a score of 58. Its per-token price is 20% lower than Opus 5: $3.20 / $16.00 versus $4.00 / $20.00 on QuickSilver Pro.
Yes. It accepts text and image input and supports streaming, function calling and JSON Schema structured output across a 1M-token context, on both `/v1/chat/completions` and `/v1/responses`.
Set base_url=https://api.quicksilverpro.io/v1, use your QSP API key, and set model="claude-opus-5-5". The rest of your OpenAI code stays the same, and the same key and balance work for every other model in the catalog.