fullauto.online

The board · maintained by hand

Everything here expires. We print the date on it.

Writing about AI tooling goes stale faster than anyone admits. So every post on this site carries a half-life meter: how old it is, and whether you should still act on it. Below is the current shape of the stack.

Models

what you build on
Claude Fable 5 Anthropic · 9 Jun 2026 Anthropic's most capable widely released model. 1M context, adaptive thinking always on, slower. $10/$50 per MTok. frontier
GPT-5.6 Sol OpenAI Frontier tier for complex professional work. 1.05M context. frontier
Claude Opus 5 Anthropic · 24 Jul 2026 What Anthropic tells you to reach for on agentic coding and enterprise work. 1M context, $5/$25. agentic
Claude Sonnet 5 Anthropic · 30 Jun 2026 The workhorse. Most agent loops should start here and move up only when they measurably need to. $2/$10. balanced
GPT-5.6 Terra OpenAI The intelligence-versus-cost middle of the GPT-5.6 line. 1.05M context. balanced
Gemini 3.6 Flash Google Current stable Flash tier, balancing speed with intelligence. Google's docs don't clearly name a 3.x Pro tier. balanced
Claude Haiku 4.5 Anthropic Cheap and fast for classification, routing and the hundred small calls inside a bigger system. 200k context, $1/$5. small
GPT-5.6 Luna OpenAI Aimed at cost-sensitive, high-volume workloads. Same 1.05M context as the rest of the line. small
Claude Mythos 5 Anthropic · Project Glasswing Fable 5's specs without the safeguards, for defensive cybersecurity. Invitation-only, no self-serve sign-up. restricted

Harnesses

what runs the loop
Claude Code Anthropic Terminal, desktop, web and IDE. The reference implementation of a coding agent with real permissions. coding
Claude Agent SDK Anthropic Claude Code's loop, tools and context management as a library. Python and TypeScript only — other languages drive the CLI as a subprocess. library
Managed Agents Anthropic Hosted REST API for long-running agents. Anthropic runs the agent and the sandbox, so you don't operate session infrastructure. hosted
Codex CLI OpenAI Sandboxed by default and open source. The usual pick when those two are hard requirements. coding
Gemini CLI Google Apache-2.0, strong on whole-repository context. The free tier ended 18 Jun 2026, so it's now a budgeted decision. coding
Editor agents Cursor, Zed, Copilot, others Agent loops living inside the editor. Tighter feedback, narrower blast radius. in-editor

Protocols & frameworks

what holds it together
MCP 2026-07-28 open spec Current spec. Stateless core, per-request capability negotiation, cacheable list results, and a formal extensions framework. protocol
MCP extensions opt-in, negotiated Tasks for long-running work with durable handles, Apps for inline UI, Skills over MCP for structured agent instructions. extensions
LangGraph LangChain Explicit state machines. Worth it when the control flow is genuinely a graph, not a loop. orchestration
Pydantic AI Pydantic Typed outputs and typed tools for people who already think in models. typed
DSPy Stanford NLP Compiles prompts against a metric instead of asking you to write them. optimizer

Hand-maintained, not a feed. Checked 12 Aug 2026 against Anthropic's model docs, OpenAI's model list, Google's Gemini docs and the MCP specification. Model names, prices and context sizes come from those pages; the framework rows are judgement, not vendor copy. Prices are per million tokens, input/output, list rate.

Latest writing

All 15 posts →