New Flagship Model Added: Claude Opus 5
Date: 2026-07-27 / (1405-05-05)
Summary
Claude Opus 5, Anthropic's new flagship Opus model, is now available on AvalAI as claude-opus-5. It combines stronger coding, knowledge work, computer use, and long-horizon agentic execution with a 1 million-token input window and configurable reasoning effort. The model supports v1/chat/completions and v1/messages fully, with partial support for v1/responses.
Details
Anthropic
We announce access to Claude Opus 5 (claude-opus-5), Anthropic's new Opus flagship for demanding software engineering, knowledge work, scientific analysis, computer use, and multi-step agentic workflows. Compared with Claude Opus 4.8, it is designed to verify its work more carefully, sustain longer tasks, and deliver more capability at the same input and output prices. Anthropic models documentation
Key Features:
- Flagship Coding and Knowledge Work: State-of-the-art results reported by Anthropic on evaluations including Frontier-Bench and GDPval-AA
- Long-Horizon Agentic Execution: Stronger at planning, using tools, verifying results, finding root causes, and iterating until a task succeeds
- 1M-Token Input Window: Up to 1,000,000 input tokens for large repositories, extensive documents, and long-running workflows
- 128K Output Capacity: Up to 128,000 output tokens for substantial code, analysis, and structured deliverables
- Adaptive Thinking: Configurable effort levels let applications trade reasoning depth against latency and token use
- Computer Use and Vision: Supports image and PDF input, computer-use workflows, and visual artifact generation
- Developer Features: Function calling, tool choice, native structured output, response schemas, prompt caching, and output configuration
- Improved Scientific Work: Anthropic reports gains over Opus 4.8 across life-science evaluations, including organic chemistry and protein-related tasks
- Alignment and Safety: Anthropic reports its lowest overall misaligned-behavior score among recent models while retaining safeguards for sensitive cyber and biology requests
- Endpoint Support: Full support on
v1/chat/completionsandv1/messages; partial support onv1/responses
Endpoint Availability
| Endpoint | Support | Notes |
|---|---|---|
v1/chat/completions | Full | OpenAI-compatible chat, reasoning, vision, structured output, and tool workflows |
v1/messages | Full | Native Anthropic-compatible messages, thinking blocks, and tool use |
v1/responses | Partial | Use supported text and basic tool features; verify required parameters before production use |
Pricing
Prices are in USD per 1 million tokens.
| Model | Input | Cached Input | Cache Creation Input | Output |
|---|---|---|---|---|
claude-opus-5 | $5.00 | $0.50 | $6.25 | $25.00 |
Claude Opus 5 keeps the $5.00 input and $25.00 output rates of Claude Opus 4.8 while reducing the cached-input rate to $0.50 per 1 million tokens. Cache creation is billed separately when prompt caching is enabled.
API Request and Response Example
Request
bash
curl https://api.avalai.ir/v1/chat/completions \
-H "Content-Type: application/json" \
-H "Authorization: Bearer $AVALAI_API_KEY" \
-d '{
"model": "claude-opus-5",
"messages": [
{
"role": "user",
"content": "Review this migration plan, identify hidden failure modes, and propose a safer staged rollout."
}
]
}'Response
The following shortened response illustrates the standard Chat Completions structure:
json
{
"id": "chatcmpl-claude-opus-5-example",
"created": 1785160800,
"model": "claude-opus-5",
"object": "chat.completion",
"system_fingerprint": null,
"choices": [
{
"finish_reason": "stop",
"index": 0,
"message": {
"content": "The main hidden risks are dual-write divergence, incompatible rollback paths, and untested consumer idempotency. Start with shadow reads, add reconciliation metrics, migrate one bounded context at a time, and require tested rollback criteria before each traffic increase.",
"role": "assistant",
"thinking_blocks": [],
"annotations": []
}
}
],
"usage": {
"completion_tokens": 68,
"prompt_tokens": 24,
"total_tokens": 92,
"completion_tokens_details": null,
"prompt_tokens_details": {
"audio_tokens": null,
"cached_tokens": 0,
"text_tokens": 24,
"image_tokens": null
}
},
"estimated_cost": {
"unit": "0.0018200000",
"irt": 278.82,
"exchange_rate": 153200
}
}SDK Usage Examples
bash
curl https://api.avalai.ir/v1/chat/completions \
-H "Content-Type: application/json" \
-H "Authorization: Bearer $AVALAI_API_KEY" \
-d '{
"model": "claude-opus-5",
"messages": [
{
"role": "user",
"content": "Analyze this repository architecture and propose a reliable modernization plan."
}
]
}'python
from openai import OpenAI
client = OpenAI(
api_key="your-avalai-api-key",
base_url="https://api.avalai.ir/v1",
)
response = client.chat.completions.create(
model="claude-opus-5",
messages=[
{
"role": "user",
"content": "Analyze this repository architecture and propose a reliable modernization plan.",
}
],
)
print(response.choices[0].message.content)javascript
import OpenAI from "openai";
const client = new OpenAI({
apiKey: process.env.AVALAI_API_KEY,
baseURL: "https://api.avalai.ir/v1",
});
const response = await client.chat.completions.create({
model: "claude-opus-5",
messages: [
{
role: "user",
content: "Analyze this repository architecture and propose a reliable modernization plan.",
},
],
});
console.log(response.choices[0].message.content);Native Anthropic SDK (v1/messages)
bash
curl https://api.avalai.ir/v1/messages \
-H "Content-Type: application/json" \
-H "x-api-key: $AVALAI_API_KEY" \
-H "anthropic-version: 2023-06-01" \
-d '{
"model": "claude-opus-5",
"max_tokens": 2048,
"messages": [
{
"role": "user",
"content": "Find the root cause of this intermittent concurrency bug and propose a test that reproduces it."
}
]
}'python
import anthropic
client = anthropic.Anthropic(
api_key="your-avalai-api-key",
base_url="https://api.avalai.ir",
)
message = client.messages.create(
model="claude-opus-5",
max_tokens=2048,
messages=[
{
"role": "user",
"content": "Find the root cause of this intermittent concurrency bug and propose a test that reproduces it.",
}
],
)
print(message.content[0].text)javascript
import Anthropic from "@anthropic-ai/sdk";
const client = new Anthropic({
apiKey: process.env.AVALAI_API_KEY,
baseURL: "https://api.avalai.ir",
});
const message = await client.messages.create({
model: "claude-opus-5",
max_tokens: 2048,
messages: [
{
role: "user",
content: "Find the root cause of this intermittent concurrency bug and propose a test that reproduces it.",
},
],
});
console.log(message.content[0].text);Adaptive Thinking and Effort
Claude Opus 5 supports adaptive thinking and configurable effort. Start with a lower effort for routine work and increase it only when evaluations show that the additional reasoning improves task success. Do not ask the model to expose hidden chain-of-thought; request a concise rationale, verification checklist, or evidence instead.
python
from openai import OpenAI
client = OpenAI(
api_key="your-avalai-api-key",
base_url="https://api.avalai.ir/v1",
)
response = client.chat.completions.create(
model="claude-opus-5",
messages=[
{
"role": "user",
"content": "Design a staged migration from a monolith to event-driven services and verify the rollback strategy.",
}
],
extra_body={
"thinking": {"type": "adaptive"},
"output_config": {"effort": "xhigh"},
},
)
print(response.choices[0].message.content)Recommended Workloads
| Workload | Why Claude Opus 5 Fits |
|---|---|
| Complex software engineering | Strong root-cause analysis, codebase understanding, careful iteration, and tool use |
| Long-running agents | Maintains task context across multi-step workflows and verifies intermediate results |
| Knowledge work | Strong performance on research, analysis, due diligence, numerical reasoning, and table work |
| Computer use | Improved performance on desktop, browser, and end-to-end business workflows |
| Scientific analysis | Gains over Opus 4.8 in life sciences, chemistry, bioinformatics, and protein-related tasks |
| Visual and interactive artifacts | Stronger visual output for interfaces, diagrams, simulations, and interactive deliverables |