DeepSeek-V3.1 Models, Grok Code Fast 1, and Enhanced Streaming Performance
Date: 2025-09-14 / (1404-06-24)
Summary
AvalAI introduces DeepSeek-V3.1 models via Azure AI across multiple endpoints, adds X.AI's new grok-code-fast-1 optimized for agentic coding workflows, and delivers significant streaming API improvements with up to 6x faster response speeds achieving 95% provider performance parity.
Details
We're excited to announce three major enhancements that significantly expand our AI capabilities and improve performance across our platform.
DeepSeek-V3.1 Models via Azure AI
We've integrated DeepSeek-V3.1, the latest advancement in reasoning AI designed for the agent era, now available through our Azure AI partnership across multiple endpoints.
DeepSeek
- deepseek-v3.1: DeepSeek-V3.1 in non-thinking mode, optimized for fast and efficient responses with hybrid inference capabilities and enhanced tool use. Documentation
Key Features:
- Context window: 128K tokens
- Max output tokens: Standard output limits
- Mode: Non-thinking mode for efficient responses
- Input pricing (cache hit): $0.07 / 1M tokens
- Input pricing (cache miss): $0.27 / 1M tokens
- Output pricing: $1.10 / 1M tokens
- Strengths: Fast responses, efficient processing, strong general capabilities, enhanced tool use
- Best for: General chat, quick responses, high-volume applications, agentic workflows
- Reasoning: Standard reasoning without exposed thinking process
- Available on:
v1/chat/completions,v1/responses,v1/messages
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="deepseek-v3.1",
messages=[
{
"role": "user",
"content": "Explain the concept of quantum entanglement in simple terms",
}
],
)
print(response.choices[0].message.content)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: "deepseek-v3.1",
messages: [
{
role: "user",
content: "Explain the concept of quantum entanglement in simple terms",
},
],
});
console.log(response.choices[0].message.content);X.AI Grok Code Fast 1
We've added X.AI's latest grok-code-fast-1, a speedy and economical reasoning model that excels at agentic coding workflows with blazing-fast inference speeds.
X.AI
- grok-code-fast-1: A purpose-built model optimized for agentic coding workflows with exceptional speed and tool mastery. Documentation
Key Features:
- Context window: Large context support for extensive codebases
- Input pricing: $0.20 / 1M tokens
- Output pricing: $1.50 / 1M tokens
- Cached input pricing: $0.02 / 1M tokens
- Strengths: Blazing fast responses, agentic coding, tool mastery (grep, terminal, file editing)
- Best for: IDE integration, code generation, debugging, pull request analysis
- Performance: 190+ tokens per second with 90%+ cache hit rates
- Languages: TypeScript, Python, Java, Rust, C++, Go
- Available on:
v1/chat/completions,v1/responses,v1/messages
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="grok-code-fast-1",
messages=[
{
"role": "user",
"content": "Create a Python function to implement binary search with proper error handling",
}
],
)
print(response.choices[0].message.content)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: "grok-code-fast-1",
messages: [
{
role: "user",
content: "Create a TypeScript interface for a user management system",
},
],
});
console.log(response.choices[0].message.content);Enhanced Streaming API Performance
We've significantly upgraded our core API infrastructure with major streaming improvements that deliver exceptional performance across all supported endpoints.
Performance Improvements:
- Speed Enhancement: Up to 6x faster streaming responses (chunks/second)
- Provider Parity: 95% identical performance to main provider streaming
- Coverage: Improvements across Chat Completions, Responses, Messages, and Gemini v1beta/models APIs
- Optimization: Advanced chunk processing and connection management
- Reliability: Enhanced error handling and connection stability
Affected Endpoints:
v1/chat/completions- Chat completions with streamingv1/responses- Response generation with streamingv1/messages- Message processing with streaming/v1beta/models- Gemini native API streaming
These improvements ensure that streaming responses now match the speed and reliability of direct provider access while maintaining all the benefits of AvalAI's unified API approach.