Agents and models, in current use

Anthropic's flagship at the time of this entry: extended thinking on, 1M-token context, the most patient of the frontier models.

The workhorse. Faster than Opus, smarter than Haiku, the model most production deployments end up on.

The fast one. Used for sub-agents, high-volume classification, and the parts of an agent loop where capability matters less than throughput.

OpenAI's reasoning-first flagship. Native chain-of-thought, three reasoning-effort tiers, the highest published benchmark scores at release.

The last non-reasoning OpenAI flagship. Better at vibes than reasoning. Quietly deprecated in favor of GPT-5.

Google's reasoning flagship. Two-million-token context, native multimodal, the only frontier model that reads PDFs without an extraction pre-pass.

The European frontier model. Codes well, refuses politely, and keeps your data in the EU.

Elon's reasoning flagship. Native Twitter/X integration, willing to discuss what other models won't.

OpenAI's peak reasoning model before GPT-5. AIME, ARC-AGI, and SWE-bench records at release.

The model that made Claude the coding default. Coding benchmarks beat GPT-4o at launch.

Google's workhorse: fast, cheap, multimodal. The Sonnet of the Gemini family.

Enterprise RAG specialist. Built for retrieval pipelines, not chat.

Anthropic's first publicly competitive release. 100K context when GPT-4 had 8K.

The model that proved 1M-token context was practical. Video and audio native.

The model that finally made the 'Claude is better than GPT-4' argument defensible.

Google's first frontier-class model, released to counter GPT-4 and Claude 3.

The omni model. Text, image, audio natively in one system. Speed doubled vs. GPT-4.

128K context, knowledge cutoff updated, cost dropped 3×. The model that made GPT-4 accessible.

The open-weights reasoning model that printed an industry shockwave. Trained at a fraction of frontier-lab costs.

Meta's fourth Llama generation. Three sizes, all-MoE, the open-weights default for serious self-hosters.

Alibaba's flagship open-weights family. Best non-English capability among open models, strong on coding.

The model that proved 7B parameters could beat 13B models. Reshaped expectations for small open models.

Microsoft's small-but-capable reasoning model. Punches above its 14B parameter count.

The model that first made the open-weights vs. closed-source comparison uncomfortable for the labs.

Anthropic's official CLI coding agent. Long-horizon refactors, tool-use harness, MCP server support.

The IDE-shaped coding agent. Tab completion, agent mode, the first $100M ARR coding tool.

Terminal-native git-aware coding agent. Bring your own API key. Read the source.

The VS Code extension that does the work in your editor — read, write, run, browse — with permission gates.

The original. Tab-completion that started the coding-agent market.

Mistral's coding specialist. 80+ programming languages, fill-in-the-middle, low latency.

The first 'fully autonomous software engineer' marketed at scale. Browser + shell + planning loop, fully cloud-hosted.

Building blocks for custom agents — tool use, prompt caching, batch API, extended thinking, memory.

OpenAI's hosted agent runtime. Threads, tools, file search, code interpreter — managed for you.

The framework that defined agent chaining before 'agent' was a product category.

OpenAI's educational multi-agent framework. Minimal, opinionated, not production-hardened.

The open protocol that standardised tool connections for LLM agents. Now an industry standard.

Google Cloud's enterprise agent platform. RAG + Gemini + enterprise controls in one product.

AWS's model marketplace. Run Claude, Llama, Titan — whatever model you need — inside your VPC.

Claude pointed at a desktop. Mouse, keyboard, screenshots. The first frontier-model computer-use API.

OpenAI's browser-only autonomous agent. Lives in a sandboxed Chrome, performs tasks, returns results.

Search-first answer engine. Citations on every claim, model-agnostic, the first credible Google alternative for fact retrieval.

Open-weights image generation flagship. Text rendering finally works.

The image generation model that made AI art a creative category. Photorealism and style control best-in-class.

Open-weights speech recognition. The transcription backbone for most AI pipelines.

AI notebook with source grounding. Generates audio summaries. No hallucination outside uploaded sources.

Inference-as-a-service for open-weights models. Fastest Llama, DeepSeek, and Mixtral access.