Anthropic

Anthropic is the AI safety company behind the Claude family of models. Founded in 2021 by former OpenAI researchers Dario and Daniela Amodei, Anthropic has built a reputation for producing models that lead on reasoning and code generation benchmarks while emphasizing safety research. On June 30, 2026 Anthropic released Claude Sonnet 5, the most agentic Sonnet-class model yet, at $2 input / $10 output per 1M tokens, launched as introductory pricing through August 31, 2026; Anthropic later cancelled the planned September 1 step to $3 / $15, so $2 / $10 is now the standard rate. Sonnet 5 ships a 1M context window with context compaction, adaptive thinking up to xhigh effort, and vendor-reported scores of 85.2 on SWE-bench Verified and 80.4 on Terminal-Bench 2.1, the latter of which actually beats Opus 4.8 on that specific benchmark. Claude Fable 5, launched June 9, 2026, sits above the family as the frontier tier: $10 / $50 per 1M tokens, 1M context with 128K output, always-on adaptive thinking, and vendor-reported scores of 80.3 on SWE-bench Pro and 85.0 on OSWorld-Verified. From June 9 through June 22 Fable 5 was included at no extra cost on Pro, Max, Team, and seat-based Enterprise plans; starting June 23 it moved to usage-credit billing on those plans, with the underlying API rate unchanged. On July 24, 2026 Anthropic shipped Claude Opus 5 at $5 / $25 per 1M tokens, holding the Opus 4.8 price and undercutting Fable 5 by exactly half, with a 1M context window as both default and maximum, 128K output, and vendor-reported scores that lead Fable 5 on OSWorld 2.0 (70.6 to 66.1), BrowseComp (90.8 to 87.4), and GDPval-AA v2 (1861 to 1747). Claude Opus 4.8, which shipped on May 28, 2026, sits below it at the same $5 / $25 pricing, with improved agentic coding (64.3% to 69.2% on Anthropic's own measure) and a fast mode roughly 2.5x quicker and three times cheaper than before. In late April 2026 the company secured up to $25 billion in additional Amazon investment (April 21) and up to $40 billion in compute and capital from Google (April 24). On May 28, 2026 Anthropic closed a $65 billion Series H at a $965 billion post-money valuation, and on June 1 it confidentially filed a draft S-1 with the SEC for a proposed IPO, getting ahead of rival OpenAI. On September 1, 2026 Anthropic shipped Claude Fable 5.1 and Claude Mythos 5.1, and held list pricing flat at $10 / $50 while cutting cache reads 75 percent to $0.25 per 1M, which the company frames as roughly 25 percent off a typical workload and up to 45 percent off heavily agentic work; both numbers depend entirely on cache hit rate. Both models carry 1M context with 128K output, Fable 5.1 adds beta mid-conversation effort adjustment, and Anthropic reports 92.6 on GPQA Diamond and 81.2 on SWE-bench Pro for Fable 5.1 without publishing a SWE-bench Verified figure. Availability is the key difference between the two: Fable 5.1 is broadly available, while Mythos 5.1 stays gated behind the Cyber Verification and Life Sciences Verification programs run in coordination with the US government, so its published rate is a list price most developers cannot transact at. On September 22, 2026 Anthropic released Claude Opus 5.5 at $4 / $20 per 1M tokens, 20 percent under Opus 5, with cache reads at $0.20 on a 0.05x multiplier, 1M context, and 128K output; Anthropic says it performs at the level of Fable 5.1 on most work, and it went straight to first on the Vals Index (69.69 percent) and the Artificial Analysis Intelligence Index (58). It is now the Claude Code default, thinking can no longer be disabled, default effort drops to medium, and Anthropic says Claude Sonnet 5.5 and Claude Haiku 5.5 will follow in the coming weeks. Opus 5 remains available and is not deprecated. Their Model Context Protocol (MCP) has become foundational infrastructure for AI agents.