LANGUAGE MODEL Anthropic Last updated:

Claude 3

Opus, Sonnet, Haiku (Tiered Frontier Family)

Anthropic's first three-tier model family — Opus (flagship, highest-quality), Sonnet (balanced), and Haiku (fast and cheap). On release, Opus was the first model to publicly outscore GPT-4 on most benchmarks, ending GPT-4's 12-month run as uncontested SOTA. Added vision input and pushed context to 200,000 tokens.

Intelligence
Below avg
Cost
High
$18.75 in / $75.00 out
Context
200K
Up to 200,000 tokens
How are Intelligence, Speed & Cost bucketed?
Intelligence and Speed buckets are percentile ranks on Artificial Analysis. Cost buckets are fixed dollar thresholds keyed off output-token price ($/M out).
Intelligence
  • Top 1%≤ 1%
  • Top 5%≤ 5%
  • Top 10%≤ 10%
  • Good≤ 25%
  • Medium≤ 50%
  • Below avg> 50%
Speed
  • Top 1%≥ 345 tok/s
  • Top 5%≥ 237 tok/s
  • Top 10%≥ 196 tok/s
  • Good≥ 146 tok/s
  • Medium≥ 90 tok/s
  • Slow< 90 tok/s
Cost
  • Freeopen weights · self-host
  • Low< $1 / M out
  • Moderate$1–5 / M out
  • High≥ $5 / M out

Why it matters

Claude 3 ended the "there's just GPT-4 and then everything else" narrative that had held for a year. Post-launch, every enterprise AI procurement discussion started including Claude as a first-tier option. The tier-based pricing is now industry standard.

Core Capabilities

Long Documents
Handles entire codebases, books, and multi-doc RAG.
Multimodal
Combines text, vision, and audio in one model.
Generative
Produces images, video, audio, or other media.
Agent Workflows
Built for tool use and autonomous tasks.

Context Window

200k tokens
≈ 154+ pages
4k Chat 聊天
32k Long docs 长文档
128k Books 整本书
400k Multi-doc 多文档
1M Codebase 整个代码库
10M
200k

Availability

API
Available
Product / App
Available
Open Source
Not released
Enterprise
Contact sales

Pricing Model

Pay per token
Input and output billed separately.
Pay-per-token

Capability / Performance

Where this model sits relative to the middle 60% of models in the tree. All scores are 0–10 (higher is better).

Lower 20% Upper 80% This model
Reasoning
AA Intelligence Index · scaled to 10
1.7
5.6
2.6
Coding
SciCode · scaled to 10
1.8
4.3
2.3
Agentic tasks
Terminal-Bench Hard · scaled to 10
0.2
3.6
2.1
Context / memory
Context window size · log-scaled
6.0
9.0
6.7
Cost efficiency
Input price ($/M tokens) · cheaper scores higher
6.2
10.0
0.2
Lower 20% 20th percentile — 20% of models score below this This model Where the current model lands Upper 80% 80th percentile — only 20% of models score above this Percentile boundaries are computed across every model in the tree that reports the underlying benchmark for each capability.

What it feels like

  • Three tiers (Opus / Sonnet / Haiku) — first time the same family covered flagship-quality and ultra-fast in one release
  • Opus Needle-in-Haystack: 99.4% recall average, 98.3% even at 200K tokens — long-context that genuinely works
  • Opus 90.5% one-shot, 89.2% zero-shot — set new state-of-the-art on multiple expert benchmarks
  • Haiku is the fastest and cheapest model in its intelligence tier as of 2024 — practical for high-volume tasks
  • Sonnet 2x faster than Claude 2 / 2.1 with higher intelligence — became the default tier
  • Opus felt notably more 'careful' than GPT-4 — safer refusals at the cost of occasional over-caution

Best use cases

  • Long-document analysis (200K context with reliable retrieval)
  • Vision + text tasks where Opus / Sonnet outperform GPT-4V
  • Deployments needing a tier-aware family (Haiku triage → Sonnet workflow → Opus hard cases)
  • Customer-facing apps where Anthropic's safety tuning is a feature, not friction

Tools to try

Not ideal for

  • Bleeding-edge reasoning by 2025 — superseded by Claude 3.5 / 3.7 / 4 family
  • Self-hosted / open-weights deployments
  • Tasks where Opus's caution/refusal patterns get in the way

Model Evolution

Claude is Anthropic's language model family.

View full evolution tree →