Agentic Chat (Claude Opus 5.5)

Anthropic released Claude Opus 5.5 on September 22, 2026, positioning it as a frontier model for long-running coding agents, research, and professional knowledge work. The new model outperforms the company's recently launched Mythos 5.1 and Fable 5.1 across agentic coding, knowledge work, and scientific reasoning benchmarks, while costing 60% less per base API token than those predecessors. The launch coincides with OpenAI introducing GPT-6 Sol and Luna, sharpening an industry race over completed work per dollar rather than raw benchmark scores. Opus 5.5 is priced at $4 per million input tokens and $20 per million output tokens, down 20% from Opus 5, with cache-read pricing cut to $0.20 per million tokens. The model posts 66.4% on Terminal-Bench 4.0, 54.4% on FrontierCode v1.1 Main, and 57.8% on CursorBench 4.0, exceeding Fable 5.1 across these tests. Early users including GitHub, Lovable, Spotify, and Optiver reported fewer tokens and steps on real coding workloads. Developers must adapt to always-on adaptive thinking, new API model ID claude-opus-5-5, and breaking changes around tool use and thinking blocks.

Anthropic has released Claude Opus 5.5, the first model in its new Claude 5.5 family, claiming performance on par with Fable 5.1 on most tasks while costing roughly 40% less to run than Opus 5. The model is available only as a managed API on the Claude Platform, AWS, Google Cloud, and Microsoft Azure, with no released weights for self-hosting. API pricing drops to $4 per million input tokens and $20 per million output tokens, with cache reads falling 60% to $0.20, and output generation runs over 30% faster than Opus 5. On benchmarks, Opus 5.5 leads on Terminal-Bench 4.0 (66.4%), OSWorld 2.0 (81.8%), and GDPval-AA, though GPT-6 Astra still tops Terminal-Bench-Science and AutomationBench. Cyber and biology requests are routed through Fable 5.1-class safeguards, thinking cannot be disabled, and a new containment test shows boundary-circumvention attempts down about 85% versus Opus 5.

Kingy AI's analysis documents Claude Opus 5.5's September 22, 2026 launch as Anthropic's new flagship general-release model, the same underlying model powering GitLab Duo's Agentic Chat. The guide confirms pricing at $4 per million input tokens and $20 per million output tokens (20% less than Opus 5), with cache reads The article also notes where Opus 5.5 loses to competitors—GPT-6 Astra leads on AutomationBench (41.4% vs 40.0%) and Terminal-Bench-Science (64.6% vs 58.7%)—and highlights breaking API changes including forced thinking that cannot be disabled, removal of forced tool choice, a default effort drop from high to medium, an

The AI×AI deep technical report covers Anthropic's September 22, 2026 launch of Claude Opus 5.5, the model family underlying GitLab Duo's Agentic Chat. It reports that Opus 5.5 achieves 89.9% on SWE-bench Pro, 93.9% on SWE Multilingual, and 61.4% on SWE-bench Multimodal, described as the highest recorded software engin The report confirms day-one deployment across the Claude Platform API, AWS Bedrock, Google Cloud Vertex AI, Microsoft Azure AI Foundry, and GitHub Copilot, and states Opus 5.5 matches or exceeds Claude Fable 5.1 across GDPval-AA v2.1, Terminal-Bench 4.0, FrontierCode v1.1, and CursorBench 4.0. The model identifier is c

BenchLM.ai's September 22, 2026 benchmark ranking places Claude Opus 5.5, the model powering GitLab Duo's Agentic Chat, as the top Anthropic model with a score of 88.5 on BenchAlign v5.6 with a 1M token context window. It leads Claude Fable 5.1 (83.3) and Claude Opus 5 (80.4), with the top three Anthropic models all fe The rankings also show Claude Opus 5.5 surpassing all previous Claude Opus variants including Opus 4.8 (70.9), Opus 4.7 Adaptive (68.8), Sonnet 5 (67.3), and Opus 4.7 (66.4), with progressively lower scores across earlier generations down to Claude 3 Opus at 28.6. This independent aggregator data confirms the substanti

Reuters reports on Anthropic's September 22, 2026 launch of Claude Opus 5.5, the underlying model behind GitLab Duo's Agentic Chat, noting it delivers performance comparable to Anthropic's top-tier Fable 5.1 while costing 40% less to run than its predecessor. The model is priced at $4 per million input tokens and $20 p The article notes Opus 5.5 underwent external testing by independent AI safety research groups Frontier Design and METR before launch, incorporating safeguards previously reserved for the most capable systems. The model is available on Amazon Web Services, Google Cloud, and Microsoft Azure platforms. Anthropic also ann

Anthropic's official Claude Platform release notes for September 22, 2026 document the launch of Claude Opus 5.5 (claude-opus-5-5), the underlying model powering GitLab Duo's Agentic Chat. The model ships with a 1M-token default context window, 128k maximum output tokens, and always-on adaptive thinking. API pricing is Additional platform-level changes introduced alongside Opus 5.5 include a Fast mode research preview available on the Claude API, and a new beta capability allowing tools to be defined inside mid-conversation system messages via the inline-tools-2026-09-15 beta header, which supports adding, removing, or migrating tool