Blog
Practical guides on AI APIs: model selection, integration, cost optimization and production tips.
Onysoft AI Gateway Node: Run AI Models on Your Own Computer and Use Them Through the API
Run open-source models like Gemma 4, Qwen3 and gpt-oss-20b on your own Windows PC or Apple Silicon Mac with Onysoft AI Gateway Node, free via the API.
Read more →Laya or Jev? We Installed the Open-Source Typed Decision Model on Our Own Server and Tested It in Turkish
We compared Apache 2.0 licensed Laya against cloud-based Jev on the same labelled Turkish support data. Changing the question format moved accuracy from 8/20 to 20/20. With install size, memory, CPU latency and cost figures.
Read more →Typed Decision Model or Chat Model? We Measured Both on a Real 446-Model Catalogue
We gave the same classification job to a typed decision model (Jev) and to a chat model with a strict JSON schema: 446 models, a 60-model controlled comparison. 72x cost difference, 4x latency, and an unexpected gap on the confidence axis.
Read more →System One Models: Getting Typed Decisions Instead of Text from AI
System One models return typed, calibrated decisions instead of prose. The noul, choice and score primitives explained with real requests, our own measurements, and how to build the same pattern on Onysoft today.
Read more →GPT-6 Astra API: Pricing, Astra Pro, Batch and Access Guide
GPT-6 Astra API guide: Onysoft pricing in USD and TRY, Astra vs Astra Pro vs half-price batch, GPT-5.6 comparison, plus Python, curl and streaming code.
Read more →Claude Fable 5.1 API: Pricing, What's New, and How to Use It
Claude Fable 5.1 API guide: effort-scaled performance, cybersecurity changes, Opus 5 and Sonnet 5 compared, pricing, and Python + curl examples.
Read more →Gemini 3.8 Flash API: Pricing, Thinking Costs, and Usage Guide
Gemini 3.8 Flash API guide: Onysoft pricing, 3.6 vs 3.7 vs 3.8 Flash, the cost of extra thinking tokens, 3.1 Pro and DeepSeek comparison, Python examples.
Read more →New AI Models September 2026: Pricing, Comparison, and API Access
New AI models of September 2026: GPT-6 Astra, Claude Fable 5.1, Gemini 3.8 Flash, DeepSeek V4.1 Flash. USD and TRY pricing, decision guide, code examples.
Read more →What Is RAG and How Do You Set It Up? Step-by-Step RAG for Turkish Data with Real Cost Math
What is RAG and how do you build it? An honest three-layer architecture — embedding, vector database, generation — with Turkish-data specifics.
Read more →Cursor API from Türkiye: Using AI Code Assistants (Cursor, Cline, Continue.dev, aider) with One Key
OpenAI-compatible API setup for Cursor, Cline, Continue.dev, and aider from Türkiye: tool-by-tool base_url steps, an honest answer on Claude Code.
Read more →AI Agent API Guide: Agent Architecture, the Tool-Calling Schema, and Loop Cost Management
How to build an AI agent over an API: the 4 layers of agent architecture, the tool-calling schema, a per-step cost table with TRY equivalents.
Read more →What Is MCP? A Model Context Protocol Guide: Architecture, Tools, and the Gateway Relationship
What is MCP (Model Context Protocol)? Host, client, server; tools, resources, prompts; how MCP differs from an LLM gateway, with code and a cost scenario.
Read more →API Key Security for LLM Applications: How Keys Leak, How to Store Them, What to Do After a Leak
API key security guide for LLM keys: the most common leak sources, safe storage with env vars and secret managers, a damage-scenario table with real prices.
Read more →Public Sector AI API Guide: LLM Access Architecture for Government Institutions and Contractors
Public sector AI API architecture in Türkiye: citizen-data masking patterns, local contract + e-invoice + simplified procurement framework.
Read more →AI API Reseller and Partner Program: Selling AI to Your Clients Under Your Own Brand and Margin
How does an AI API reseller program work? Partner portal mechanics (per-client keys, your own margin, client-level reports), an illustrative revenue.
Read more →AI APIs in Financial Services: KVKK, Data Boundaries, and Masking (Banks, Insurers, Fintechs)
An AI API guide for banks, insurers, and fintechs: the KVKK counterparty question, a worked data-masking pattern, an honest on-prem vs API analysis.
Read more →E-Commerce AI API Guide: Product Descriptions, Chatbots, and Recommendations (with Transparent TRY Cost Math)
Which AI model fits each e-commerce job? Cost math for 10,000 product descriptions, working Python/curl code, Turkish quality notes, and KVKK masking.
Read more →Call Center AI Integration: Summaries, Intent Classification, Real Latency Data, and an Open Cost Model
AI in the call center: conversation summaries, intent classification, live reply drafts, and quality analysis. Real latency from 6,959 production requests.
Read more →Enterprise AI API Procurement and e-Invoicing in Türkiye: an End-to-End Guide for Purchasing and Accounting
How enterprise AI API purchases work in Türkiye: the e-Fatura/e-Arşiv flow with live GİB VKN lookup, the KDV-2/withholding difference.
Read more →AI API Cost Control: Six Mechanisms That Make Surprise Bills Technically Impossible
AI API cost control guide: pre-request balance checks (×1.2 buffer, 402), per-key token limits, the cost field in every response, PDF usage reports.
Read more →GPT vs Claude vs Gemini 2026: GPT-6 Astra vs Claude Fable 5.1
GPT vs Claude vs Gemini: GPT-6 Astra, Claude Fable 5.1 and Gemini 3.8 Flash on coding, writing, speed and cost, with Onysoft prices and a decision matrix.
Read more →Let AI Answer the Phone: 24/7 Voice Assistant Architecture and Real Latency Data
How to let AI answer your company phone: layer-by-layer architecture, real latency measured over 4,553 requests, an open cost breakdown and compliance notes.
Read more →One API for Multiple AI Models: Single Key, Single Balance, Single Bill
How to use multiple LLMs through one API: 750+ models behind a single key and a single invoice, real 30-day usage rankings.
Read more →Türkiye LLM Gateway Guide: What an LLM Gateway Is, Real Latency Data, and How to Choose
What an LLM gateway is and why Türkiye changes the equation: live price table with TRY equivalents, real latency from 6,959 production requests.
Read more →Türkiye's AI Action Plan 2026-2030: What It Means for Developers and Companies
Türkiye's AI Action Plan 2026-2030 explained for developers and companies: four axes, 2,000+ public datasets, a 1 GW data center target, TÜBİTAK's Bilge model, and KVKK-aware first steps.
Read more →Claude Opus 5 API: Pricing, Adaptive Thinking, and Usage Guide (Python + curl)
Claude Opus 5 API guide: 1M-token context, 128K output, adaptive thinking by default, Opus 5 vs Sonnet 5 vs Fable 5, Python, curl and streaming examples.
Read more →OnyRouter: Automatic AI Model Selection Per Request (onysoft/auto API Guide)
OnyRouter analyzes every request and routes it to the best AI model automatically: just set the model to onysoft/auto. Routing is free.
Read more →The Cheapest AI APIs in 2026: A Practical Cost-Reduction Guide
Which AI API is cheapest in 2026? Economy model classes, five cost-cutting tactics, hidden fees, and why a Turkish Lira balance keeps budgets predictable.
Read more →Is There a Free AI API? An Honest Guide for 2026
Is there a free AI API? The honest answer: nothing unlimited exists. What free tiers really offer, their hidden costs, and how to start for pennies instead.
Read more →Gemini API in Turkey: Free Tier Limits, Pricing, and Paying in Lira (2026)
Gemini API from Turkey: free tier limits, access through Onysoft, and Gemini 3.6 Flash and 3.1 Pro with a lira balance and e-invoices.
Read more →Grok API in Turkey: Access, Paying in Turkish Lira, and Usage (2026)
Use the xAI Grok API from Turkey: Turkish Lira balance at central-bank rates, corporate e-invoices, and Grok 4.5, 4.3, and 4.20 through the same OpenAI SDK.
Read more →DeepSeek API in Turkey: The Most Economical Coding Model, Paid in Lira (2026)
Use DeepSeek V4 and V3.2 from Turkey with a lira balance: a strong, budget-friendly coding model behind one OpenAI-compatible key.
Read more →Kimi K3 API in Turkey: Moonshot's Flagship, Paid in Turkish Lira (2026)
How teams in Turkey access the Kimi K3 API: skip foreign cards and USD billing — pay in Turkish Lira at central bank rates, get e-invoices.
Read more →AI Models Guide 2026: GPT-6 Astra, Claude Fable 5.1, Gemini 3.8
Which AI model fits which job? GPT-6 Astra, Claude Fable 5.1, Gemini 3.8 Flash, Grok 4.6, Qwen3.8 and DeepSeek V4.1, with a decision tree and Onysoft prices.
Read more →Claude API in Turkey 2026: Live Prices in Lira, Access, and Usage Guide
A practical 2026 guide to the Claude API in Turkey: a live USD + Turkish Lira price table for Opus 5, Sonnet 5, and Haiku 4.5 at the central-bank rate.
Read more →Kimi K3 API: Access and Usage Guide (Python + curl)
Kimi K3 API access guide: Moonshot AI's open-weight flagship, 1M-token context, benchmark standing vs Claude and GPT, plus Python and curl examples.
Read more →Gemini 3.6 Flash API: Pricing, Speed, and Usage Guide (Python + curl)
Gemini 3.6 Flash API guide: 1M-token context, lower pricing than 3.5 Flash, a head-to-head comparison, plus Python, curl, and streaming examples.
Read more →Accessing the ChatGPT / OpenAI API from Turkey: Paying in Turkish Lira (2026 Guide)
A practical guide for teams in Turkey: use the OpenAI API with a Turkish Lira balance, central-bank exchange rates, corporate e-invoices.
Read more →AI API Pricing 2026: Live Price Table in USD and Turkish Lira, With Real Cost Scenarios
AI API pricing 2026: live USD + Turkish lira prices for 12 models (TCMB rate, Sep 13, 2026) and real monthly cost math for a chatbot and a coding assistant.
Read more →KVKK-Compliant AI API Usage: Turkey's 2025-2026 AI Guidance and an Enterprise Compliance Checklist
A practical guide to KVKK-compliant AI API integration in Turkey: the regulator's 2025-2026 generative AI, agentic AI and workplace guidance.
Read more →Grok 4.5 API: Access and Usage Guide (Python + curl)
Grok 4.5 API guide: the 500K context window, Grok 4.5 vs 4.3 vs 4.20, the ~x-ai/grok-latest alias, and Python and curl examples via one API.
Read more →GPT-5.6 Family (Sol, Terra, Luna): API Access and Model Choice Guide
GPT-5.6 API guide: Sol vs Terra vs Luna tiers, -pro variants, 1M+ token context, migrating from GPT-5.5, and Python and curl examples via one API.
Read more →Claude Fable 5 API: Access and Usage Guide
Claude Fable 5 API access guide: Mythos-class model, 1M context, Fable 5 vs Opus 4.8 vs Sonnet 5, Python and curl examples via one OpenAI-compatible API.
Read more →OpenAI SDK base_url Setup for Türkiye: Switch to 750+ Models with One Line
How to change base_url in the OpenAI SDK: one-line switch in Python, Node.js and curl, the OPENAI_BASE_URL environment-variable method, an SDK version table.
Read more →Gemini 3.1 Pro API: What Can You Actually Do With a 1M-Token Context Window?
What a 1M-token context window enables in practice: whole-codebase review, long-document analysis, RAG-free prompting with Gemini 3.1 Pro. Python example inside.
Read more →How to Reduce LLM API Costs: A Practical Model Selection Guide
A practical guide to reducing LLM API costs: match tasks to model tiers, use cascade routing, trim tokens, and track price/performance across 750+ models.
Read more →Veo 3.1, Kling 3.0, Runway Aleph: AI Video Generation Through a Single API
Generate AI video with Veo 3.1, Kling 3.0 and Runway Aleph via one OpenAI-compatible API: async task flow, curl/Python examples, prompt tips, and cost logic.
Read more →