New Models Added: Stable Imagen 4.0, DeepSeek-V3.1, and BFL FLUX Models

Date: 2025-08-24

Summary

We're excited to announce major model updates including stable versions of Google's Imagen 4.0 series, DeepSeek-V3.1 with enhanced reasoning capabilities, and new FLUX models from Black Forest Labs (BFL). These additions provide improved stability, advanced reasoning, and cutting-edge image generation and editing capabilities.

---

Details

Google Imagen 4.0 - Stable Versions Now Available

The stable versions of Google's Imagen 4.0 series are now available, replacing the previous preview versions with enhanced reliability and performance:

• imagen-4.0-generate-001: High-quality image generation with improved stability and consistency. Supports resolutions up to 2048x2048 and multiple aspect ratios. Documentation
• imagen-4.0-fast-generate-001: Optimized for speed while maintaining quality, perfect for applications requiring quick image generation. Documentation
• imagen-4.0-ultra-generate-001: Ultra-high quality image generation with exceptional detail and realism. Supports the highest resolution outputs up to 2816x1536. Documentation

Key Features of Imagen 4.0 Stable Models:

• Digital watermarking and verification
• User-configurable safety settings
• Prompt enhancement using prompt rewriter
• Person generation capabilities
• Multiple aspect ratios and high-resolution support

DeepSeek-V3.1 - Enhanced Reasoning and Agent Capabilities

We're announcing support for DeepSeek-V3.1, the latest model from DeepSeek.com that brings significant improvements in reasoning and agent capabilities. The existing model names deepseek-chat and deepseek-reasoner now automatically route to DeepSeek-V3.1:

• deepseek-chat: Non-thinking mode of DeepSeek-V3.1 for fast, efficient responses
• deepseek-reasoner: Thinking mode of DeepSeek-V3.1 with advanced reasoning capabilities

DeepSeek-V3.1 Improvements:

• Hybrid inference: Think & Non-Think modes in one model
• Faster thinking: Reaches answers quicker than previous versions
• Stronger agent skills: Enhanced tool use and multi-step agent tasks
• 128K context: Extended context window for both modes
• Better performance: Improved results on SWE-bench and Terminal-Bench
• Efficient reasoning: Significant gains in thinking efficiency

Important Note: Only deepseek-chat and deepseek-reasoner route to the new DeepSeek-V3.1. Other model names like deepseek-r1-0528 and deepseek-v3-0324 use different providers (Azure and AWS Bedrock) and do not resolve to DeepSeek-V3.1.

BFL FLUX Models - Advanced Image Generation and Editing

New FLUX models from Black Forest Labs are now available, offering state-of-the-art image generation and editing capabilities:

• flux-1.1-pro: Advanced image generation model supporting v1/images/generations endpoint with superior performance across image quality, prompt adherence, and generation speed. Documentation
• flux.1-kontext-pro: Versatile model supporting both v1/images/generations and v1/images/edits endpoints, excelling in text editing and character preservation tasks. Documentation

Flux Models Key Features:

• High Performance: Top rankings in image quality and prompt adherence benchmarks
• Fast Generation: Consistently lower latencies than competing models
• Versatile Editing: Advanced text editing and character preservation (flux.1-kontext-pro)
• Multiple Endpoints: Support for both generation and editing workflows

Usage Examples

Google Imagen 4.0 Models

Python

from openai import OpenAI

client = OpenAI(api_key="your-avalai-api-key", base_url="https://api.avalai.ir/v1")

response = client.images.generate(
    model="imagen-4.0-generate-001",
    prompt="A serene mountain landscape with a crystal-clear lake reflecting the peaks at golden hour",
    size="1024x1024",
    n=1,
    response_format="url",
)

print(response.data[0].url)

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.images.generate({
 model: "imagen-4.0-generate-001",
 prompt: "A serene mountain landscape with a crystal-clear lake reflecting the peaks at golden hour",
 size: "1024x1024",
 n: 1,
 response_format: "url", // or b64_json
});

console.log(response.data[0].url);

Bash

curl https://api.avalai.ir/v1/images/generations \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $AVALAI_API_KEY" \
  -d '{
 "model": "imagen-4.0-generate-001",
 "prompt": "A serene mountain landscape with a crystal-clear lake reflecting the peaks at golden hour",
 "size": "1024x1024",
 "n": 1,
 "response_format": "url"
 }'

DeepSeek-V3.1 Models

Python

from openai import OpenAI

client = OpenAI(api_key="your-avalai-api-key", base_url="https://api.avalai.ir/v1")

# Using deepseek-chat (non-thinking mode)
response = client.chat.completions.create(
    model="deepseek-chat",
    messages=[
        {
            "role": "user",
            "content": "Explain the concept of quantum entanglement in simple terms",
        }
    ],
)

print(response.choices[0].message.content)

# Using deepseek-reasoner (thinking mode)
reasoning_response = client.chat.completions.create(
    model="deepseek-reasoner",
    messages=[
        {
            "role": "user",
            "content": "Solve this complex math problem: If a train travels 120 miles in 2 hours, then speeds up by 25% for the next 3 hours, how far did it travel in total?",
        }
    ],
)

print(reasoning_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",
});

// Using deepseek-chat (non-thinking mode)
const response = await client.chat.completions.create({
    model: "deepseek-chat",
    messages: [
        {
            role: "user",
            content: "Explain the concept of quantum entanglement in simple terms",
        },
    ],
});

console.log(response.choices[0].message.content);

// Using deepseek-reasoner (thinking mode)
const reasoningResponse = await client.chat.completions.create({
    model: "deepseek-reasoner",
    messages: [
        {
            role: "user",
            content: "Solve this complex math problem: If a train travels 120 miles in 2 hours, then speeds up by 25% for the next 3 hours, how far did it travel in total?",
        },
    ],
});

console.log(reasoningResponse.choices[0].message.content);

BFL (FLUX) Models - Image Generation

Python

from openai import OpenAI

client = OpenAI(api_key="your-avalai-api-key", base_url="https://api.avalai.ir/v1")

response = client.images.generate(
    model="flux-1.1-pro",
    prompt="A futuristic cityscape at sunset with flying cars and neon lights",
    size="1024x1024",
    n=1,
    response_format="b64_json",  # not supporting 'url'
)

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.images.generate({
 model: "flux-1.1-pro",
 prompt: "A futuristic cityscape at sunset with flying cars and neon lights",
 size: "1024x1024",
 n: 1,
 response_format: "b64_json", // not supporting 'url'
});

Bash

curl https://api.avalai.ir/v1/images/generations \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $AVALAI_API_KEY" \
  -d '{
 "model": "flux-1.1-pro",
 "prompt": "A futuristic cityscape at sunset with flying cars and neon lights",
 "size": "1024x1024",
 "n": 1,
 "response_format": "url"
 }'

BFL (FLUX) Models - Image Editing

Python

from openai import OpenAI

client = OpenAI(api_key="your-avalai-api-key", base_url="https://api.avalai.ir/v1")

response = client.images.edit(
    model="flux.1-kontext-pro",
    image=open("original_image.png", "rb"),
    prompt="Change the text on the sign to 'Welcome to AvalAI'",
    size="1024x1024",
    n=1,
    response_format="b64_json",  # not supporting 'url'
)

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.images.edit({
 model: "flux.1-kontext-pro",
 image: fs.createReadStream("original_image.png"),
 prompt: "Change the text on the sign to 'Welcome to AvalAI'",
 size: "1024x1024",
 n: 1,
 response_format: "b64_json", // not supporting 'url'
});

Bash

curl https://api.avalai.ir/v1/images/edits \
  -H "Authorization: Bearer $AVALAI_API_KEY" \
  -F image="@original_image.png" \
  -F model="flux.1-kontext-pro" \
  -F prompt="Change the text on the sign to 'Welcome to AvalAI'" \
  -F size="1024x1024" \
  -F n=1

Pricing Information

The new models are available with competitive pricing:

• imagen-4.0-generate-001: $0.04 per image
• imagen-4.0-fast-generate-001: $0.02 per image
• imagen-4.0-ultra-generate-001: $0.06 per image
• flux-1.1-pro: $0.04 per image
• flux.1-kontext-pro: $0.04 per image
• DeepSeek-V3.1 models:
  • deepseek-chat: $0.07 / 1M tokens (cache hit), $0.27 / 1M tokens (cache miss), $1.10 / 1M tokens (output)
  • deepseek-reasoner: $0.14 / 1M tokens (cache hit), $0.55 / 1M tokens (cache miss), $2.19 / 1M tokens (output)
  • Pricing reference: DeepSeek API Pricing

---

Related Links

• Google Imagen Documentation
• DeepSeek Documentation
• BFL (FLUX) Documentation
• Image Generation Guide
• Chat Completions API Reference
• Images API Reference
• Model Selection Guide
• Rate Limits and Pricing