Frontier

Compare flagship models side by side across context, pricing, endpoint support, regional availability, and agent harness compatibility.

9 models

Emerging

2 models

Claude Fable 5.1 Published Sep 1 ’26

Context: 1M
Max output: 128K
Knowledge cutoff: June 2026
Input modalities

Pricing

In: $10.00
Out: $50.00
Cache Read: $ 0.25
Cache Write: $12.50

Endpoint APIs

bedrock-runtime: ConverseResponsesInvoke
bedrock-mantle: ConverseResponsesInvoke

Regional availability

Global
Geo: US
In-Region: 1 regions

Features

Tool calling: —
Structured outputs

GPT-6 Astra Published Sep 8 ’26

Context: 1M
Max output: 128K
Knowledge cutoff: April 30, 2026
Input modalities

Pricing

In: $10.00 (short)$20.00 (long)
Out: $50.00 (short)$75.00 (long)
Cache Read: $ 1.00 (short)$ 2.00 (long)
Cache Write: $12.50 (short)$25.00 (long)

Endpoint APIs

bedrock-runtime: ConverseResponsesInvoke
bedrock-mantle: ConverseResponsesInvoke

Regional availability

Global
Geo: US
In-Region: 2 regions

Features

Tool calling
Structured outputs

Flagship

3 models

Claude Opus 5.5 Published Sep 22 ’26

Context: 1M
Max output: 128K
Knowledge cutoff: June 2026
Input modalities

Pricing

In: $ 4.00
Out: $20.00
Cache Read: $ 0.20
Cache Write: $ 5.00

Endpoint APIs

bedrock-runtime: ConverseResponsesInvoke
bedrock-mantle: ConverseResponsesInvoke

Regional availability

Global
Geo: US · EU · AU · JP
In-Region: 4 regions

Features

Tool calling
Structured outputs

GPT-6.1 Sol Published Sep 29 ’26

Context: 1M
Max output: 131.1K
Knowledge cutoff: Not documented
Input modalities

Pricing

In: $ 2.00 (short)$ 4.00 (long)
Out: $10.00 (short)$15.00 (long)
Cache Read: $ 0.10 (short)$ 0.20 (long)
Cache Write: $ 2.50 (short)$ 5.00 (long)

Endpoint APIs

bedrock-runtime: ConverseResponsesInvoke
bedrock-mantle: ConverseResponsesInvoke

Regional availability

Global
Geo: US
In-Region: 1 regions

Features

Tool calling: —
Structured outputs

Grok 4.7 Published Sep 28 ’26

Context: 524,288 tokens, shared between input and output
Max output: 524,288 minus input
Knowledge cutoff: Not documented
Input modalities

Pricing

In: $ 2.00
Out: $ 6.00
Cache Read: $ 0.50
Cache Write: —

Endpoint APIs

bedrock-runtime: ConverseResponsesInvoke
bedrock-mantle: ConverseResponsesInvoke

Regional availability

Global
Geo: US
In-Region

Features

Tool calling
Structured outputs

Balanced

2 models

Claude Sonnet 5.5 Published Sep 28 ’26

Context: 1M
Max output: 128K
Knowledge cutoff: June 2026
Input modalities

Pricing

In: $ 2.00
Out: $10.00
Cache Read: $ 0.10
Cache Write: $ 2.50

Endpoint APIs

bedrock-runtime: ConverseResponsesInvoke
bedrock-mantle: ConverseResponsesInvoke

Regional availability

Global
Geo: US · EU · AU
In-Region: 1 regions

Features

Tool calling
Structured outputs

Grok 4.3 Published Jun 15 ’26

Context: 1M
Max output: —
Knowledge cutoff: Not documented
Input modalities

Pricing

In: $ 1.25
Out: $ 2.50
Cache Read: $ 0.20
Cache Write: —

Endpoint APIs

bedrock-runtime: ConverseResponsesInvoke
bedrock-mantle: ConverseResponsesInvoke

Regional availability

Global
Geo
In-Region: 4 regions

Features

Tool calling
Structured outputs

Fast

2 models

Claude Haiku 4.5 Published Oct 16 ’25

Context: 200K
Max output: 64K
Knowledge cutoff: Feb 2025
Input modalities

Pricing

In: $ 1.00
Out: $ 5.00
Cache Read: $ 0.10
Cache Write: $ 1.25

Endpoint APIs

bedrock-runtime: ConverseResponsesInvoke
bedrock-mantle: ConverseResponsesInvoke

Regional availability

Global
Geo: US · EU · AU · JP · IN
In-Region: 8 regions

Features

Tool calling
Structured outputs

GPT-6 Luna Published Sep 22 ’26

Context: 1M
Max output: 128K
Knowledge cutoff: May 18, 2026
Input modalities

Pricing

In: $ 0.10 (short)$ 0.20 (long)
Out: $ 0.50 (short)$ 0.75 (long)
Cache Read: $ 0.01 (short)$ 0.02 (long)
Cache Write: $ 0.125 (short)$ 0.25 (long)

Endpoint APIs

bedrock-runtime: ConverseResponsesInvoke
bedrock-mantle: ConverseResponsesInvoke

Regional availability

Global
Geo: US
In-Region: 1 regions

Features

Tool calling
Structured outputs