Model

Claude Instant 1.2

CLOSED

Fast, low-cost tier that seeded the later Haiku lineage; 100K context, improved math/coding/safety over v1.1.

ReasoningCodingVisionFunction callingTool useAgentic

558.7

Elo · rank #286

Parameters

Undisclosed

Active params

—

Context

100K tokens

Architecture

Dense decoder-only Transformer (parameter count undisclosed)

License

Proprietary (Anthropic Commercial Terms of Service)

Languages

—

API price (in/out)

$1.63 / $5.51

Modalities

text

Benchmark results

Bar shows position within the tracked field; marker = field best

GSM8KMath86.7%#57

best: Llama 3.1 405B · 96.8%

HumanEvalCoding58.7%#104

best: Claude Opus 4.5 · 99.4%

MATH-500Math49.9%#133

best: GPT-5 · 99.4%

MMLUKnowledge63.1%#190

best: OpenAI o3 · 92.9%

OpenBookQAReasoning84.4%#6

best: Claude 1 · 90.8%

Run it locally

Closed weights — available via API only. No local deployment.

Input / M tok

$1.63

Output / M tok

$5.51

Claude Instant family

Elo progression across releases

Resources

API price $1.63/$5.51 · each benchmark row carries its own source badge (see methodology)