Claude Sonnet 4.5

claude-sonnet-4-5anthropic

Claude Sonnet 4.5 is served through QVeris by anthropic. Compare its current input formats, context limits and usage price below to evaluate it for your application.

Input format: image, text → text
Usage pricing: 1,575 / 7,875
Context tokens: 200,000
Maximum output tokens: 64,000

Overview

ChatStreamingTool callingVision

What to evaluate

These evaluation ideas follow the capabilities enabled here. Test representative tasks before choosing a model.

Tool-assisted agents and multi-step automation
Image interpretation and visual questions
Long documents and codebase context

API

Call this model with your QVeris API key. The examples below use this model’s public ID and the API endpoint for this site.

Base URL: https://aigateway.qveris.ai/v1
Model ID: claude-sonnet-4-5

1. Create an API key

Keep your API key on the server in the QVERIS_API_KEY environment variable. Do not expose it in browser code. Get an API key →

2. Send a request

Endpoint

curl -sS 'https://aigateway.qveris.ai/v1/chat/completions' \
  -H "Authorization: Bearer $QVERIS_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
  "model": "claude-sonnet-4-5",
  "messages": [
    {
      "role": "user",
      "content": "Hello!"
    }
  ],
  "max_tokens": 64
}'

Run these examples on your server. JavaScript uses Node.js 18+; Python uses the standard library.

For streaming chat, set stream to true. Read SSE data until [DONE]; the qveris.billing event contains settlement details.

Successful requests are billed in Credits at the prices shown below. Save the X-QVeris-Call-ID response header for billing and support.

API integration

Usage pricing

Prices are the Credits charged by QVeris. Credits purchased in different packages may have different monetary costs.

Input1,575Credits / 1M tokens

Output7,875Credits / 1M tokens

Cached input157.5Credits / 1M tokens

Rates vary with total input length. All rates below are Credits per 1M tokens.
Input tokens from	Input	Output	Cached input
0	1,575	7,875	157.5
200,001	3,150	11,812.5	315

Final charges follow actual usage and the price effective when the request starts. Cached-input rates apply only to cache hits.

Credits and top-up pricing

Estimate a request

Indicative estimate in Credits for one request, not a quote. Input includes cached tokens. Reasoning tokens, tools or other meters may add usage; final settlement follows the effective rate.

Total input tokensOutput tokensCached input tokens

Estimated Credits: 5.5125

Capabilities and limits

Context tokens: 200,000
Maximum output tokens: 64,000
Input modalities: image, text
Output modalities: text
Supported API operations: chat.completions messages responses

Integration notes

Only the capabilities listed above are advertised by this serving route. Optional provider parameters and agent features may require additional validation.

The context window includes input and generated output. Keep room for the response; tokenization and reasoning usage can differ by model.

Performance depends on the serving route and request. No gateway latency or uptime measurements are published here yet.

Model information and sources

A model-specific source has not been added yet. The capability and price information above is supplied by QVeris.

More models from this provider

Claude Fable 5
Claude Fable 5 is served through QVeris by anthropic. Compare its current input formats, context limits and usage price below to evaluate it for your application.
Claude Fable 5.1
Claude Fable 5.1 is served through QVeris by anthropic. Compare its current input formats, context limits and usage price below to evaluate it for your application.
Claude Fable 5.2
Claude Fable 5.2, released by Anthropic on November 9, 2026, is designed for demanding reasoning, coding, visual understanding, and long-horizon agentic work. It supports a 1M-token context window and up to 128,000 output tokens. Adaptive thinking is always on, with low, medium, high, xhigh, and max effort; the API default is high. Supports tools, structured outputs, and streaming. Standard USD/MTok pricing: input 8, output 40, cache reads 0.20, 5m cache writes 10, 1h cache writes 16. Omit temperature, top_p, and top_k; forced tool choices any/tool are unsupported.
Claude Haiku 5.5
Claude Haiku 5.5 is served through QVeris by anthropic. Compare its current input formats, context limits and usage price below to evaluate it for your application.