The Open-Source AI Stack
RSS
All models

Models · claude

Claude Sonnet 4

Proprietary Anthropic · 2025-05-22 · Proprietary

Mid-tier model of the May 22 2025 Claude 4 launch. Inherited the hybrid-reasoning approach from Claude 3.7 Sonnet with near-instant and extended-thinking modes, plus parallel tool execution and an extended-thinking-with-tool-use beta. Held the SWE-Bench Verified lead for closed mid-tier coding through summer 2025 at the same $3 / $15 price point as 3.5 and 3.7 Sonnet.

Cost

$3.00 / Mtok input
$15.00 / Mtok output

Anthropic API · as of 2026-05-21

source ↗

Speed

48.4 tok/sec output
966 ms TTFT

· as of 2026-05-21

source ↗

Architecture

tokens in Embedding vocab not disclosed × N layers Architecture not disclosed (proprietary or undocumented) Output projection tokens out
Schema-generated from data/models.yaml. Every label is auditable against the model's sources.

Specs

Architecture
unknown
Total params
not disclosed
Active params
not disclosed
Context window
not verified
Attention
unknown
Position encoding
unknown
Post-training
rlhf, constitutional
OSI-approved
no
Data released
no
Training code
not released

Benchmarks

Each score carries the date it was published; we never infer or interpolate missing scores.

General reasoning

MMLU-Pro 83.7 as of 2026-05-21 source ↗
GPQA-Diamond 70.0 as of 2025-05-22 source ↗

Code

SWE-Bench Verified 72.7 as of 2025-05-22 source ↗
LiveCodeBench 44.9 as of 2026-05-21 source ↗

Math

MATH 93.4 as of 2026-05-21 source ↗
AIME 2024 40.7 as of 2026-05-21 source ↗
AIME 2025 33.1 as of 2025-05-22 source ↗

Available quantizations

None. The weights are not distributed, so there are no public quantizations.

Notable innovations

  • · Hybrid reasoning with tool use in extended thinking
  • · Parallel tool execution
  • · 65% reduction in reward-hacking shortcuts vs Sonnet 3.7

Lineage

Continues the Sonnet line: 3.5 Sonnet to 3.7 Sonnet to Sonnet 4 to Sonnet 4.5.

Derived from

Claude 3.7 Sonnet 2025-02-24

Derivatives

Sources