Claude Opus 5.5 Launch: #1 on AI Index, 20% Price Cut (2026)
Claude Opus 5.5 is Anthropic's newest AI model, released on September 22, 2026, and the first in its Claude 5.5 family. If ChatGPT is your go-to AI, check the new scoreboard: Claude Opus 5.5 scored 58 on the independent Artificial Analysis Intelligence Index, the highest mark that index has recorded.
What Changed in Claude Opus 5.5?
Anthropic says Opus 5.5 matches its top model, Fable 5.1, on most work. Prices fall 20% to $4 per million input tokens and $20 per million output, typical workloads cost 40% less, and output is over 30% faster. It is also the first release since CEO Dario Amodei called for pacing the frontier; METR and other evaluators tested it first.
Claude Opus 5.5 vs GPT-6 Astra: Benchmark Results
On Anthropic's own table, Opus 5.5 beats OpenAI's GPT-6 Astra on agentic coding (Terminal-Bench 4.0, 66.4% vs 57.9%), Humanity's Last Exam with tools (67.7% vs 57.2%) and real-world work (GDPval-AA, 1,846 vs 1,542). Astra leads on automation (41.4% vs 40.0%) and science tasks (64.6% vs 58.7%). The coding runs used different settings, and independent testing scored both models at 59.6%.
On the nine-model ranking, Opus 5.5 (58) leads Fable 5.1 and Astra (53), while Grok 4.7 scores 46 and Gemini 3.8 Flash 41.
OpenAI's GPT-6 Sol and Luna, Plus Hermes Agent on Claude
About 90 minutes after Anthropic's launch, OpenAI released GPT-6 Sol for coding and agents (AI that completes multi-step tasks on its own) and GPT-6 Luna for high-volume jobs, both at half the price of the GPT-5.6 series. Two days earlier, Nous Research shipped an official, still-experimental plugin that runs its open-source Hermes Agent on Claude Pro and Max plans.
What Claude Opus 5.5 Means for the AI Market
Two price cuts in one day show the race shifting from raw intelligence to value per task. OpenAI now sells three GPT-6 tiers, and Claude Sonnet 5.5 and Haiku 5.5 follow within weeks. Analysts expect more vendors to meter AI agents separately, which would make choosing a model more like picking a plan for US enterprises.
Key Takeaways
① #1 on the index - Claude Opus 5.5 scored 58 on the independent index, ahead of GPT-6 Astra's 53.
② 20% cheaper - $4 in and $20 out per million tokens, with output over 30% faster.
③ Price war - OpenAI answered in 90 minutes with half-price GPT-6 Sol and Luna.
Scores still matter, but the next round will be fought on price.
👉 Claude Fable 5.1 Launch: 2x GPT-5.6 on Science, 75% Cache Cut (2026) - also worth a read.
📌 Sources: Anthropic, OpenAI, TechCrunch, Artificial Analysis, Nous Research, InfoWorld (2026)
댓글
댓글 쓰기