Claude Opus 5.5: What Is New and How It Compares
By ChatGBot · September 29, 2026
On September 22, 2026, Anthropic released Claude Opus 5.5, the first model in the new Claude 5.5 family. It is both stronger and cheaper than the model it replaces, and it arrived on the same day OpenAI shipped its own new tiers, kicking off a real price war. Here is what changed and whether it matters for how you work.
Every top AI in one app
GPT, Claude, DeepSeek, Qwen and GLM — one subscription.
Try ChatGBot →Cheaper than Opus 5, and faster to run
Opus 5.5 is priced at $4 per million input tokens and $20 per million output tokens, about 20% below Opus 5 at $5 / $25. Cached input drops to $0.20. Anthropic says typical workloads run roughly 40% cheaper because the model reaches an answer with less back and forth, not just because the sticker price is lower. For background on the lineup, see Claude models explained and our Claude Opus guide.
Better at agentic work
The headline improvement is agentic performance: multi step tasks where the model plans, calls tools, and follows a long chain of work without losing the thread. Anthropic says Opus 5.5 beats the previous flagship on these benchmarks, which is exactly the kind of work coding assistants and autonomous agents lean on. If you use Claude for building things, this is the upgrade that shows up in daily use. See Claude for coding.
The first of the Claude 5.5 family
Opus is the top tier, but it is not the whole story. Anthropic has said Sonnet 5.5 and Haiku 5.5 are expected in the following weeks, which usually means a mid tier that balances speed and cost, and a small, fast tier for high volume tasks. If you want the cheapest option today, the smaller models are worth waiting for.
Opus 5.5 vs GPT-6: pick by the task
About ninety minutes after Opus 5.5, OpenAI launched GPT-6 Sol ($2 / $10) and GPT-6 Luna ($0.10 / $0.50). The takeaway is not that one model won. It is that the frontier is now an argument about cost per task. Reach for Opus 5.5 or GPT-6 Sol on hard reasoning and coding, and a cheap model like GPT-6 Luna, DeepSeek or GLM for summaries and bulk work. The easiest way to do that is to keep them side by side and switch as the task changes.
Frequently asked questions
When was Claude Opus 5.5 released? September 22, 2026. It is the first model in the Claude 5.5 family.
How much does Claude Opus 5.5 cost? $4 per million input tokens and $20 per million output tokens, about 20% below Opus 5, with cached input at $0.20.
Is Claude Opus 5.5 better than Opus 5? Anthropic says it beats the previous flagship on agentic benchmarks while running typical workloads about 40% cheaper.
Can I use Claude Opus in ChatGBot? Yes. ChatGBot gives you Claude along with GPT, DeepSeek, Qwen and GLM under one subscription, so you can switch models mid chat.