ModelPriceLabModelPriceLab
Changes/Claude Opus 4.6 introduced (1M context beta, 128k output, $5/$25 per MTok)
MajorAnthropicNew Launchclaude-opus-4-6Verified

Claude Opus 4.6 introduced (1M context beta, 128k output, $5/$25 per MTok)

Anthropic introduced Claude Opus 4.6 with improved coding, long-horizon agentic work, and a 1M token context window (beta). Standard API pricing remains $5/$25 per million input/output tokens, with premium long-context pricing above 200K input tokens.

Feb 5, 2026Views:16Effective: Feb 5, 2026

Impact Assessment

Cost0
Reliability2
Migration0
Quality5
Compliance0
What changed
  • New model id: claude-opus-4-6
  • 1M token context window (beta) with premium pricing when input > 200K tokens
  • Up to 128k output tokens
  • New API controls: adaptive thinking + effort levels; context compaction (beta)
Pricing (API)
  • Standard: $5 / MTok input, $25 / MTok output
  • Long-context (>200K input tokens with 1M context enabled): $10 / MTok input, $37.50 / MTok output
  • Batch: $2.50 / MTok input, $12.50 / MTok output
Recommended actions
  • Validate code generation/refactor quality and long-running agent behavior on your repo.
  • If you plan to use 1M context, budget for premium long-context rates and verify your org eligibility (tier 4 / custom limits).

Changes

No diff data available.

Recommended Actions

1

TEST

Test Opus 4.6 on your main coding workflows (generation, refactor, review, debugging), and compare against your current baseline model.

2

MONITOR

If enabling 1M context, add cost guardrails for long-context premium pricing and validate org eligibility.

Sources

blog
Introducing Claude Opus 4.6

Pricing remains the same at $5/$25 per million tokens; Opus 4.6 adds 1M context (beta) and 128k output tokens.

docs
Claude API Pricing

Long context pricing: Opus 4.6 standard $5/$25 per MTok; premium long-context above 200K input tokens $10/$37.50.

Comments

Loading comments...