PickModel

Claude Haiku 4.5

Anthropic

Lite

Anthropic's fastest and most compact model. Ideal for lightweight agents, high-throughput pipelines, and latency-sensitive applications.

Claude Opus 4.7

Anthropic

Frontier

Anthropic's most capable model. Exceptionally strong at complex reasoning, research, coding, and agentic tasks. Sets new standards for frontier AI.

Claude Sonnet 4.6

Anthropic

Standard

Anthropic's best model for coding and everyday tasks. Exceptional balance of speed, intelligence, and cost — the workhorse of the Claude family.

DeepSeek R1

DeepSeek

Frontier

DeepSeek's open-source reasoning model. Matches o1-level math and coding with full transparency — the thinking chain is visible in the output.

DeepSeek V3

DeepSeek

Frontier

Chinese lab DeepSeek's impressive open-source flagship. Competitive with top-tier models at a dramatically lower inference cost.

DeepSeek V4

NEW

DeepSeek

Frontier

DeepSeek's latest flagship MoE model with 1.6T parameters and 49B active. Dramatically expands context to 1M tokens with optional thinking mode — still at a fraction of the cost of Western labs.

Gemini 2.5 Flash

Google

Standard

Google's workhorse model — fast, cheap, and surprisingly capable. Handles 1M token contexts at a fraction of Pro's cost with optional thinking mode.

Gemini 3.1 Pro

Google

Frontier

Google's current flagship. Leads coding and reasoning benchmarks with a 2M-token context window, native multimodal inputs, and agentic optimisations.

GLM-5.1

NEW

Z.ai

Frontier

Z.ai's frontier reasoning model built on a 744B MoE architecture. Tops SWE-Bench Pro coding benchmarks and excels at agentic tasks at a fraction of the price of comparable Western models.

GPT-4o

OpenAI

Frontier

OpenAI's flagship omnimodal model. Handles text, audio, and images natively in a single model, with strong reasoning and broad multilingual support.

GPT-4o mini

OpenAI

Lite

OpenAI's affordable small model with vision. Replaces GPT-3.5 Turbo with stronger performance and multimodal capabilities at a fraction of the cost.

Grok 3

xAI

Frontier

xAI's latest flagship model trained on a massive compute cluster. Strong math and science capabilities with real-time web search integration.

Llama 3.3 70B

Meta

Standard

Meta's high-performance open-source model. Near frontier quality in a 70B package — fully free to download, self-host, and fine-tune.

Llama 4 Maverick

Meta

Frontier

Meta's latest multimodal open model. A mixture-of-experts architecture delivers frontier-class performance in an open, self-hostable package.

MiMo-V2.5-Pro

NEW

Xiaomi

Frontier

Xiaomi's flagship agentic model with 1T parameters and full multimodal support. Tops SWE-bench Pro and ClawEval at 8x lower output cost than Claude Opus.

MiniMax M2.7

NEW

MiniMax

Standard

MiniMax's compact 230B MoE reasoning model with only 10B active parameters. Punches well above its weight on intelligence benchmarks with fast inference and two API speed tiers.

Mistral Large 2

Mistral AI

Standard

Mistral's flagship model with strong multilingual and coding abilities. Available under a permissive license and deployable on-premise.

o3

OpenAI

Frontier

OpenAI's most powerful reasoning model. Excels at science, math, coding, and complex multi-step problems through deep chain-of-thought reasoning.

o4-mini

OpenAI

Standard

A compact reasoning model from OpenAI. Delivers strong math and coding performance at much lower cost than o3, with faster response times.

Qwen3.6 Plus

NEW

Alibaba

Frontier

Alibaba's latest hybrid MoE model with always-on chain-of-thought reasoning and a 1M-token context window. Competitive with frontier models on coding and agentic benchmarks at a very low price point.