Claude

Claude is the family of large language models made by Anthropic. Anthropic released it on March 14, 2023, after a closed alpha with partners, in two versions: Claude and Claude Instant, a lighter, less expensive and faster option.¹ Today’s models come in named lines. As of October 13, 2026 the current lineup is Claude Fable 5.1, for demanding reasoning and long-horizon agentic work; Claude Opus 5.5, for long-running agentic coding and knowledge work; Claude Sonnet 5.5; and Claude Haiku 5.5, for high-volume, latency-sensitive tasks. All four accept text and image input, have a 1M-token context window and up to 128K output tokens, and Anthropic gives June 2026 as their reliable knowledge cutoff.² Anthropic notes that the tokenizer introduced with Claude Opus 4.7 produces about 30% more tokens for the same text than its earlier one, which matters when comparing per-token prices.³

On ARC-AGI-3’s Semi-Private set, run by ARC Prize on its Standard harness, Claude Opus 5 scores 30.16%, third among 19 models, and Claude Opus 4.8 1.52% (Highest ARC-AGI-3 Scores, by Model).⁴ Claude 4.5 Opus has the top SWE-bench Verified score in the shared mini-SWE-agent v2.0.0 harness, 76.8%, and Opus 5.5 running in Claude Code tops the Terminal-Bench 4.0 leaderboard at 64.85% over 330 trials (SWE-bench Verified, Pro and What Replaced Them).⁵⁶ List prices run from $0.50 per 1M output tokens for Haiku 5.5, for prompts up to 100,000 tokens, to $50 for Fable 5.1 (Cheapest LLM APIs per 1M Output Tokens).³ On OpenRouter, Claude Opus 5.5 processed 5.35 trillion tokens in the week to October 12, 2026, eighth among all models, and Claude Sonnet 5.5 2.65 trillion (Most-Used AI Models on OpenRouter by Weekly Tokens).⁷