DeepSeek small agent model: 284B/13B MoE, 1M context, MIT weights, $0.14/$0.28 per 1M tokens.

DeepSeek flagship 1.6T MoE with 49B active, 1M context, MIT weights. API $0.29/$1.20 cached $0.03.

Alibaba Qwen3.8 2.4T MoE with 95B active, 1M context, multimodal input. Coder $0.14/$0.80; 397B $0.29/$1.20.

OpenAI GPT-5.6 Luna: fast affordable GPT-5.6 tier, 128k context, $1/$6 per 1M tokens. Rechecked 2026-08-18.

OpenAI GPT-5.6 Terra: balanced daily driver in the 5.6 family, 400k context, $2.50/$15 per 1M tokens.

OpenAI frontier flagship GPT-5.6 Sol: 1.05M context, effort up to max plus ultra, $5/$30 per 1M tokens.

Manus is an action engine for slides, sites, design, research, and browser tasks. Official banner: it will soon operate as an independent company again.

SpaceXAI's frontier 1.5T reasoning model with 500K-token context, agentic coding strength, and real-time X data at roughly half the price of rivals.

Zhipu coding model on GLM Coding Plan. Rechecked: Lite $10, Pro $30, Max $60. Token API not on the public table yet. Not $18-only.

Replit Agent 4 builds apps, sites, decks, and mobile from plain language on a hosted workspace, with a visual canvas and parallel agents.

Anthropic's Mythos-class flagship made safe for general use, with 1M-token context, frontier coding and vision, days-long agentic autonomy, and $10/$50 per 1M tokens pricing.

Anthropic's frontier Opus model with 1M-token context and effort-controlled reasoning, near-Fable-5 intelligence at Opus-4.8 pricing for agents and coding.

Moonshot AI's open-weight 2.8T multimodal agentic model with 1M-token context, the world's first open 3T-class model rivaling closed frontier models.

Persistent, searchable memory for AI coding agents with 95.2% recall on LongMemEval, running locally on SQLite with MCP, hooks, and 54 tools for Claude Code, Cursor, and Codex.

ByteDance's open-source multimodal AI agent stack for controlling computers and browsers, with Agent TARS CLI, remote operators, and the UI-TARS-1.5 model.

Anthropic's enterprise code security scanning, now powered by Claude Mythos 5 in public beta, tracing data across files to report CWE categories, severity, and suggested fixes.

Cognition's local AI IDE (Windsurf 2.0 / Devin Desktop). Plan on your machine, one-click handoff to Devin in the cloud, review the PR in the same app.

Google's most capable Flash model for agentic workflows and multimodal reasoning, with 1M-token context, 65k output tokens, and Computer Use preview.