Skip to content

models.md

Latest commit

 

History

History
103 lines (67 loc) · 4.37 KB

models.md

File metadata and controls

103 lines (67 loc) · 4.37 KB
Raw

Models — Specifications & Pricing

← Back to README

The Claude 4 model family at a glance, plus deeper specs for the current flagship. For Anthropic's authoritative list (always more current), see the official model overview.

The Claude 4 family

Model Model ID Context Max output Best for
Opus 4.7 (flagship) claude-opus-4-7 200K (1M beta via API) 128K Complex reasoning, large refactors, multi-file analysis
Sonnet 4.6 (balanced) claude-sonnet-4-6 200K 8K Everyday coding — most tasks live here
Haiku 4.5 (fast/cheap) claude-haiku-4-5-20251001 200K 8K Quick edits, doc updates, light reasoning

⚠️ Anthropic ships new models often. Always check the official model overview before pinning a model ID.

Previous-generation models still available: claude-opus-4-6 and claude-sonnet-4-5-20250929 remain accessible via API and the /model switcher. They're no longer the default but useful for cost optimization or pinning a build to a specific behavior.

Claude Opus 4.7 — deep specs

Key specifications

Feature Specification
Model ID claude-opus-4-7
Context window 200K tokens (1M beta via API)
Max output 128K tokens
Availability Pro, Max, and API plans
Fast Mode Available via Opus 4.6 (/fast switches the underlying model — see Fast Mode below)

What's new in 4.7 vs 4.6

  • Sharper adaptive thinking. Reasoning depth scales more reliably with task complexity — fewer cases of overthinking trivial prompts or underthinking hard ones.
  • Better long-context grounding. Multi-file refactoring at the upper end of the 200K window holds the thread tighter than 4.6.
  • Improved tool calling. Lower rate of malformed JSON and clearer reasoning when picking between similar tools.
  • Same price. Input/output pricing matches Opus 4.6.

Capabilities

1. Extended context window — 200K standard for everyone; 1M beta via API only. Best for: analyzing entire large codebases at once, processing extensive documentation.

2. Adaptive thinking — dynamic reasoning that scales with task complexity. You can hint at depth with "think", "think hard", "think harder", or "ultrathink" in your prompt; for the persistent dial, see Effort levels. Opus 4.7 defaults to the xhigh effort level on every plan.

3. Agent teams (experimental) — multi-agent collaboration in one session. Enable with CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1.

4. Top-tier benchmarks — leading scores on Terminal-Bench 2.0 and SWE-bench Verified.

When to reach for Opus 4.7

  • Complex architectural decisions
  • Large-scale refactoring or migrations
  • Debugging intricate, multi-file issues
  • Production-critical code paths

Save tokens with Sonnet 4.6 or Haiku 4.5 when:

  • The task is a single-file edit or simple refactor
  • You're writing or updating documentation
  • You're doing exploratory Q&A about an unfamiliar codebase

Fast Mode

Fast Mode delivers ~2.5× faster responses at 6× the standard token price. It runs on Opus 4.6 — not 4.7. Toggle with /fast; the ↯ indicator confirms it's on. Available via API and Pro/Max plans.

Standard Opus 4.6 Fast Mode (Opus 4.6)
Input (per MTok) $5 $30 (6×)
Output (per MTok) $25 $150 (6×)

💡 Default to Opus 4.7 for everyday work. Reach for /fast only when latency genuinely matters (live debugging, demo prep, time-pressured fixes).

Pricing

Subscription plans

Plan Price Models Usage
Pro $20/mo ($17/mo annual) All ~45 messages / 5hr
Max 5x $100/mo All 5× Pro
Max 20x $200/mo All 20× Pro

API (pay-as-you-go)

Model Input (per MTok) Output (per MTok)
Opus 4.7 $5 $25
Sonnet 4.6 $3 $15
Haiku 4.5 $1 $5
Opus 4.6 (Fast Mode only) $30 $150

Pricing changes occasionally. Check anthropic.com/pricing before committing.

← Back to README · FAQ · Changelog