- 1M context
- Multimodal
- Responses API
Claude Haiku 5.5 on QuickSilver Pro
Claude Haiku 5.5: $0.025 input / $0.125 output per 1M tokens, 75% below Anthropic list ($0.10 / $0.50). Context: 1M tokens. Maximum output: 128K tokens. Available through Chat Completions and Responses. Capabilities: text input, text output, image input, streaming and tool calling.
- 1M context
- 128K max output
- Vision
- Tools
- Streaming
- No JSON Schema output
You pay · input / output per 1M
$0.025 / $0.125
Prompts over 200K tokens: $0.125 / $0.625
75% lower
on both legs vs Anthropic list price
Pricing comparison ($/1M tokens)
Bar length = price relative to the list price
| Provider | Input | Output | vs QSP |
|---|---|---|---|
| QuickSilver Pro | $0.025 | $0.125 | lowest-cost |
| Anthropic list price(Anthropic API) | $0.10 | $0.50 | 75% lower |
Anthropic list source: anthropic.com/pricing
At a glance
- Context
- 1M tokens
- Input / 1M
- $0.025
- Output / 1M
- $0.125
- Thinks by default
- No
- Max output
- 128K
- Latency · live
- —
For developers building subagents and high-volume chat or extraction workflows.
When is Claude Haiku 5.5 the right call?
Curated from catalog capabilities and published prices, not a benchmark.
When to use
Use Claude Haiku 5.5 for subagents and high-volume workflows. Context: 1M tokens; maximum output: 128K tokens. The gateway drops temperature and top_p and rewrites assistant prefill as a user continuation.
- Long documents and codebases, up to 1M tokens of context
- Agents that call tools
- Prompts that include images
- Responses API clients as well as Chat Completions
When to use something else
For reasoning support, use Claude Sonnet 5.5: 1M context / 128K maximum output tokens at 20× this model’s base output price. Use Claude Haiku 4.5 when you need temperature or assistant prefill without a rewrite; its context is 200K tokens. Above 200K prompt tokens, this model bills the whole request at $0.125 input / $0.625 output per million tokens.
Quickstart (curl)
Get an API key →curl https://api.quicksilverpro.io/v1/chat/completions \
-H "Authorization: Bearer $QSP_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "claude-haiku-5-5",
"messages": [{"role": "user", "content": "Hello!"}]
}'import os
from openai import OpenAI
client = OpenAI(
base_url="https://api.quicksilverpro.io/v1",
api_key=os.environ["QSP_KEY"],
)
response = client.chat.completions.create(
model="claude-haiku-5-5",
messages=[{"role": "user", "content": "Hello!"}],
)
print(response.choices[0].message.content)import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://api.quicksilverpro.io/v1",
apiKey: process.env.QSP_KEY,
});
const response = await client.chat.completions.create({
model: "claude-haiku-5-5",
messages: [{ role: "user", content: "Hello!" }],
});
console.log(response.choices[0].message.content);curl https://api.quicksilverpro.io/v1/responses \
-H "Authorization: Bearer $QSP_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "claude-haiku-5-5",
"input": "Hello!"
}'import os
from openai import OpenAI
client = OpenAI(
base_url="https://api.quicksilverpro.io/v1",
api_key=os.environ["QSP_KEY"],
)
response = client.chat.completions.create(
model="claude-haiku-5-5",
messages=[{"role": "user", "content": "Hello!"}],
)
print(response.choices[0].message.content)OpenAI-compatible. One-line migration via base_url.
Gateway behavior
Compatibility notes for this model.
temperature, top_p- Dropped before the request is forwarded.
assistant prefill- This model rejects a trailing assistant prefill, so the gateway rewrites it as a user continuation turn carrying the prefix.
tool_choice (forced)- Retained on this model: a forced tool choice can require that tool call.
image_url (remote)- Public remote URLs are fetched and inlined. An inaccessible or rejected URL fails the request; inline base64 is also supported.
prompt- Prompts up to 1M tokens. Above 200K prompt tokens the request bills at $0.125 input / $0.625 output per 1M.
max_tokens- Up to 128K output tokens.
- /v1/chat/completions: Supported
- /v1/responses: Supported
- streaming: Supported
- tools: Supported
- image input: Supported
- JSON Schema: Not listed for this model
- reasoning: Not listed for this model
Where the price comes from
- Provider reference
- $0.10 / $0.50anthropic.com/pricing
- Catalog verified
- Prompt cache · not compared
- $0.0025 read · $0.03125 5m write · $0.05 1h write
- Long-context tier
- $0.125 / $0.625 over 200K
usage.cost = prompt_tokens × $0.025/1M + completion_tokens × $0.125/1M
FAQ
Common questions
Claude Haiku 5.5 costs $0.025 input / $0.125 output per 1M tokens. Anthropic list is $0.10 / $0.50; the discount is 75% on both base rates.
For Claude Haiku 5.5, prompts above 200K tokens use $0.125 input / $0.625 output per 1M tokens for the request. Below or at that threshold, base rates are $0.025 / $0.125. The 1M-token context limit is separate from the billing threshold.
The gateway drops temperature and top_p. This model rejects a trailing assistant prefill, so the gateway rewrites it as a user continuation turn carrying the prefix.
The gateway retains a forced tool_choice on this model. Claude response_format with json_schema is advisory. Use a tool with strict: true for enforced argument shape; forced tool choice can require that tool call.
Catalog capabilities: text input, text output, image input, streaming and tool calling. For Claude image requests, the gateway fetches public remote image URLs and inlines the image data. An inaccessible or rejected URL fails the request; inline base64 is also supported.
Set base_url="https://api.quicksilverpro.io/v1", use your QSP API key, and set model="claude-haiku-5-5". Supported APIs: Chat Completions and Responses. See the runnable quickstart on this page and the API docs.