Anthropic Opus 4.6 Launch Adds Multi-Agent Teams at $350B Valuation

Zach Anderson Feb 05, 2026 18:43

Anthropic releases Claude Opus 4.6 with 1M token context window and agent teams feature, days after $350B tender offer valuation shook tech stocks.

Anthropic Opus 4.6 Launch Adds Multi-Agent Teams at $350B Valuation

Anthropic dropped Claude Opus 4.6 on Thursday, a model upgrade that quintuples its context window to one million tokens and introduces autonomous multi-agent collaboration—arriving just one day after the company's $350 billion tender offer valuation triggered a selloff across tech stocks.

The timing matters. Investors spooked by AI competition dumped shares on February 4, and now Anthropic is showing exactly why it commands that valuation: Opus 4.6 outperforms OpenAI's GPT-5.2 by 144 Elo points on GDPval-AA, a benchmark measuring economically valuable knowledge work in finance, legal, and technical domains.

What Actually Changed

Three upgrades stand out for enterprise users.

The 1M token context window (in beta) represents a 5x increase from Opus 4.5's 200,000 tokens. On MRCR v2—a needle-in-a-haystack retrieval test—Opus 4.6 scores 76% compared to Sonnet 4.5's 18.5%. That's not incremental improvement; that's a different capability class for document-heavy workflows.

Agent Teams in Claude Code lets developers spin up multiple AI agents working in parallel. Early access partner Invariant Labs reported Opus 4.6 "autonomously closed 13 issues and assigned 12 issues to the right team members in a single day, managing a ~50-person organization across 6 repositories." The model handled both product and organizational decisions while knowing when to escalate to humans.

For enterprise integration, Anthropic added PowerPoint support (research preview) alongside upgraded Excel capabilities. The model can now ingest unstructured data, infer structure without guidance, and execute multi-step changes in one pass.

Benchmark Performance

Opus 4.6 leads on Terminal-Bench 2.0 for agentic coding and tops Humanity's Last Exam, a multidisciplinary reasoning test. It also beat every other model on OpenAI's BrowseComp, which measures ability to locate hard-to-find information online.

The 190 Elo point improvement over its own predecessor on GDPval-AA suggests meaningful progress on tasks that actually generate revenue—financial analysis, legal review, technical documentation.

Pricing holds steady at $5/$25 per million tokens for input/output. Premium pricing ($10/$37.50) applies for prompts exceeding 200k tokens.

Safety Claims and Competitive Positioning

Anthropic says Opus 4.6 shows "the lowest rate of over-refusals of any recent Claude model" while maintaining alignment comparable to Opus 4.5. The company added six new cybersecurity probes to detect potential misuse, acknowledging the model's enhanced capabilities cut both ways.

The release positions Anthropic directly against OpenAI in the enterprise AI race. With Google's Gemini and xAI's Grok also competing for enterprise contracts, Opus 4.6's multi-agent features and extended context window represent Anthropic's bet that autonomous AI workflows—not just chatbots—will define the next phase of enterprise adoption.

Opus 4.6 is available immediately on claude.ai, the API, and major cloud platforms. Developers can access it via the claude-opus-4-6 model identifier.

Image source: Shutterstock