The model library

Anthropic

Anthropic's cheapest and fastest Claude: a big step up from Haiku 4.5 in coding, computer use and knowledge work, at $0.10 / $0.50 per 1M tokens for prompts up to 100K.

Added 7 Oct 2026 · text

Mistral

Mistral's 1.05T-parameter mixture-of-experts flagship (49B active), text and image in, served from the EU. Public preview at half the list price.

Added 6 Oct 2026 · text

Google

Google DeepMind's new frontier model for long-horizon software engineering, legal and finance work, and cybersecurity defence, with a 1M-token output limit. Trusted testers only for now.

Announced 30 Sep 2026 · text

Anthropic

The fast Claude 5.5: near-Opus coding and knowledge work, 30%+ faster than Sonnet 5, at half Opus 5.5's price.

Added 28 Sep 2026 · text

OpenAI

OpenAI's April 2025 small model with a million-token window, still running inside a lot of production agents. Keep it as your baseline and test newer models against it on the same key.

Added 27 Sep 2026 · text

Z.ai

Z.ai's flagship coding and agent model, GLM-5.2's base post-trained further, at the same price.

Added 23 Sep 2026 · text

Xiaomi

The previous MiMo flagship, same rate card as V2.6-Pro, a million-token reasoning model under a dollar.

Added 23 Sep 2026 · text

Meta

Meta Superintelligence Labs' frontier model, fourth on the Artificial Analysis index, served by Meta directly.

Added 23 Sep 2026 · text

Anthropic

First of the Claude 5.5 generation: Fable-class agentic coding and computer use at 20% below Opus 5.

Added 22 Sep 2026 · text

OpenAI

GPT-6's high-volume tier: the cheapest frontier-family model in the catalog, with a million-token window.

Added 22 Sep 2026 · text

OpenAI

GPT-6 for complex coding and agent work at a fifth of Astra's price, with the same million-token window.

Added 22 Sep 2026 · text

Xiaomi

Xiaomi's trillion-parameter flagship: frontier-class agentic coding on the price/intelligence Pareto line.

Added 16 Sep 2026 · text

Google

Google's cheapest text model here, for high-volume classification and extraction.

Added 11 Sep 2026 · text

Google

Cost-efficient Gemini for high-volume agentic work, translation and simple data processing.

Added 11 Sep 2026 · text

Google

The previous-generation Flash, balancing speed and multimodal work across everyday agentic tasks.

Added 11 Sep 2026 · text

Google

Google's most capable Flash model, built for long-horizon software engineering and autonomous agents.

Added 11 Sep 2026 · text

DeepSeek

DeepSeek's cheapest current model, a million-token window at flash rates.

Added 10 Sep 2026 · text

Z.ai

Native multimodal coding model at the lowest rate in the catalog.

Added 10 Sep 2026 · text

OpenAI

OpenAI's GPT-6 flagship, a million-token window one key away.

Added 3 Sep 2026 · text

Anthropic

Anthropic's top generally available model, tied for first on the Artificial Analysis Intelligence Index.

Added 1 Sep 2026 · text

Google

The Gemini 3 Flash preview line.

Added 30 Aug 2026 · text

Google

Google's newest Pro tier, shipped as a preview.

Added 30 Aug 2026 · text

Google

The previous stable Flash, still current for most work.

Added 30 Aug 2026 · text

Google

Google's current fast model, million-token context.

Added 30 Aug 2026 · text

xAI

Million-token context at roughly half the flagship rate.

Added 30 Aug 2026 · text

xAI

The previous flagship, same rate card.

Added 30 Aug 2026 · text

xAI

xAI's flagship, half-million-token context.

Added 30 Aug 2026 · text

Anthropic

Anthropic's first Mythos-class model, the new flagship tier above Opus, tuned for the hardest reasoning and agentic work.

Added 5 Aug 2026 · text

Anthropic

The Claude 5 generation's deep-work model, maximum-capability reasoning for research, architecture, and hard agentic tasks.

Added 5 Aug 2026 · text

Z.ai

Z.ai's open-weight frontier challenger, strong coding and reasoning with one of the best free tiers in class.

Added 1 Aug 2026 · text

DeepSeek

Retired by DeepSeek on 14 Sep 2026. Requests are served by V4.1 Flash from then, and billed at Flash's rate.

Added 15 Jul 2026 · text

Moonshot AI

Moonshot's value tier, K3-style agentic tool-calling at half the price, a favorite for high-volume agent loops.

Added 20 Jun 2026 · text

Anthropic

Fast, cheap frontier-family model for high-volume tasks.

Added 1 Jun 2026 · text

Anthropic

The frontier tier for the hardest reasoning and agentic work.

Added 1 Jun 2026 · text

Anthropic

Frontier-quality text and reasoning through one OpenAI-compatible endpoint.

Added 1 Jun 2026 · text

OpenAI

OpenAI's flagship, one key away.

Added 1 Jun 2026 · text

Moonshot AI

Agentic coding model with strong multi-step tool use.

Added 1 Jun 2026 · code

Mistral

Mistral's flagship 128B dense model, chat, reasoning and coding merged into one, served from the EU.

Added 28 Apr 2026 · text

Perplexity

Perplexity's decision model: typed answers with probabilities, text or image state, $0.04 per million input tokens and free output.

Added 1 Oct 2026 · decision

TypeSafe

Typed decisions with calibrated probabilities in about 100 ms. No generated text, so nothing to parse and nothing to hallucinate.

Added 15 Sep 2026 · decision