—: Data not disclosed. Prices and performance may come from different API routes; measurement settings and sources appear in the technical details.
Model technical details and sources
01
Anthropic
Claude Fable 5.2 (max)
Anthropic's Fable release for technical tasks, coding and visual reasoning. It follows Fable 5.1; access terms and usage limits may vary by provider.
AA intelligence score
65
Context
1,000,000 token
Input / output
$8 / $40 / 1M token
Generation speed
95 tokens/s
Time to first token
249.03 s
Output limit
128,000 token
Published
2026-11-09
License
Özel kullanım koşulları
Input types
Text, Image
Prices are in USD per 1 million tokens for this API route on OpenRouter. Context and output limits apply to this route, not to all services from the developer. The catalog date has not been used as the model release date. Search and other tool calls may incur additional charges.
Evaluation: Speed and latency: Artificial Analysis, Claude Fable 5.2 (max). Time to the first answer token may include thinking time. Prices come from a separate OpenRouter route, which should not be assumed to match the measurement route.
Artificial Analysis Intelligence Index v5.0.1. Model configuration in the source: Claude Fable 5.2 (max). This is an index calculated from multiple tests, not a success percentage. The source was checked on November 12, 2026; this has not been used as the test date.
Google's Gemini 4 model for coding, agent tasks and complex reasoning. It can process text, image, audio and video inputs.
AA intelligence score
60
Context
1,048,576 token
Input / output
$2 / $10 / 1M token
Generation speed
77 tokens/s
Time to first token
65.84 s
Output limit
943,718 token
Published
Not disclosed
License
Özel kullanım koşulları
Input types
Audio, Image, Text, Video
Prices are in USD per 1 million tokens for this API route on OpenRouter. Context and output limits apply to this route, not to all services from the developer. The catalog date has not been used as the model release date. Search and other tool calls may incur additional charges.
Evaluation: Speed and latency: Artificial Analysis, Gemini 4 Carbon (high). Time to the first answer token may include thinking time. Prices come from a separate OpenRouter route, which should not be assumed to match the measurement route.
Artificial Analysis Intelligence Index v5.0.1. Model configuration in the source: Gemini 4 Carbon (high). This is an index calculated from multiple tests, not a success percentage. The source was checked on November 12, 2026; this has not been used as the test date.
An Opus release for changes to large codebases and long agent tasks. Aimed at code review and multistep reasoning.
AA intelligence score
59
Context
1,000,000 token
Input / output
$4 / $20 / 1M token
Generation speed
115 tokens/s
Time to first token
353.58 s
Output limit
128,000 token
Published
Not disclosed
License
Özel kullanım koşulları
Input types
Text, Image
Prices are in USD per 1 million tokens for this API route on OpenRouter. Context and output limits apply to this route, not to all services from the developer. The catalog date has not been used as the model release date. Search and other tool calls may incur additional charges.
Evaluation: Speed and latency: Artificial Analysis, Claude Opus 5.5 (max). Time to the first answer token may include thinking time. Prices come from a separate OpenRouter route, which should not be assumed to match the measurement route.
Artificial Analysis Intelligence Index v5.0.1. Model configuration in the source: Claude Opus 5.5 (max). This is an index calculated from multiple tests, not a success percentage. The source was checked on November 12, 2026; this has not been used as the test date.
The upper-tier GPT-6-family model for advanced analysis, software development and research. It focuses on agent workflows that use tools during long tasks.
AA intelligence score
57
Context
1,050,000 token
Input / output
$10 / $50 / 1M token
Generation speed
38 tokens/s
Time to first token
254.53 s
Output limit
128,000 token
Published
Not disclosed
License
Özel kullanım koşulları
Input types
Image, Text
Prices are in USD per 1 million tokens for this API route on OpenRouter. Context and output limits apply to this route, not to all services from the developer. The catalog date has not been used as the model release date. Different pricing thresholds apply to long inputs; current thresholds appear in the linked pricing table. Search and other tool calls may incur additional charges.
Evaluation: Speed and latency: Artificial Analysis, GPT-6 Astra (max). Time to the first answer token may include thinking time. Prices come from a separate OpenRouter route, which should not be assumed to match the measurement route.
Artificial Analysis Intelligence Index v5.0.1. Model configuration in the source: GPT-6 Astra (max). This is an index calculated from multiple tests, not a success percentage. The source was checked on November 12, 2026; this has not been used as the test date.
A Sonnet release for feature development, bug fixes and everyday work documents. It processes text and images to produce text responses.
AA intelligence score
55
Context
1,000,000 token
Input / output
$2 / $10 / 1M token
Generation speed
145 tokens/s
Time to first token
254.06 s
Output limit
128,000 token
Published
Not disclosed
License
Özel kullanım koşulları
Input types
Text, Image
Prices are in USD per 1 million tokens for this API route on OpenRouter. Context and output limits apply to this route, not to all services from the developer. The catalog date has not been used as the model release date. Search and other tool calls may incur additional charges.
Evaluation: Speed and latency: Artificial Analysis, Claude Sonnet 5.5 (max). Time to the first answer token may include thinking time. Prices come from a separate OpenRouter route, which should not be assumed to match the measurement route.
Artificial Analysis Intelligence Index v5.0.1. Model configuration in the source: Claude Sonnet 5.5 (max). This is an index calculated from multiple tests, not a success percentage. The source was checked on November 12, 2026; this has not been used as the test date.
An updated GPT-6 Sol version for coding agents, computer use and document-heavy work. It generates text responses from text and images.
AA intelligence score
55
Context
1,050,000 token
Input / output
$2 / $10 / 1M token
Generation speed
55 tokens/s
Time to first token
166.79 s
Output limit
128,000 token
Published
Not disclosed
License
Özel kullanım koşulları
Input types
Image, Text
Prices are in USD per 1 million tokens for this API route on OpenRouter. Context and output limits apply to this route, not to all services from the developer. The catalog date has not been used as the model release date. Different pricing thresholds apply to long inputs; current thresholds appear in the linked pricing table. Search and other tool calls may incur additional charges.
Evaluation: Speed and latency: Artificial Analysis, GPT-6.1 Sol (max). Time to the first answer token may include thinking time. Prices come from a separate OpenRouter route, which should not be assumed to match the measurement route.
Artificial Analysis Intelligence Index v5.0.1. Model configuration in the source: GPT-6.1 Sol (max). This is an index calculated from multiple tests, not a success percentage. The source was checked on November 12, 2026; this has not been used as the test date.
The GPT-6.1-family model for everyday coding, writing and agent tasks. It can interpret screenshots and is listed at a lower token price than the Sol tier.
AA intelligence score
51
Context
1,050,000 token
Input / output
$1 / $5 / 1M token
Generation speed
73 tokens/s
Time to first token
165.25 s
Output limit
128,000 token
Published
Not disclosed
License
Özel kullanım koşulları
Input types
Image, Text
Prices are in USD per 1 million tokens for this API route on OpenRouter. Context and output limits apply to this route, not to all services from the developer. The catalog date has not been used as the model release date. Different pricing thresholds apply to long inputs; current thresholds appear in the linked pricing table. Search and other tool calls may incur additional charges.
Evaluation: Speed and latency: Artificial Analysis, GPT-6.1 Terra (max). Time to the first answer token may include thinking time. Prices come from a separate OpenRouter route, which should not be assumed to match the measurement route.
Artificial Analysis Intelligence Index v5.0.1. Model configuration in the source: GPT-6.1 Terra (max). This is an index calculated from multiple tests, not a success percentage. The source was checked on November 12, 2026; this has not been used as the test date.
A Gemini model for software development, math and scientific problem solving. Offered for agent workflows that carry context across multiple steps.
AA intelligence score
50
Context
1,048,576 token
Input / output
$2 / $10 / 1M token
Generation speed
65 tokens/s
Time to first token
39.53 s
Output limit
943,718 token
Published
Not disclosed
License
Özel kullanım koşulları
Input types
Audio, Image, Text, Video
Prices are in USD per 1 million tokens for this API route on OpenRouter. Context and output limits apply to this route, not to all services from the developer. The catalog date has not been used as the model release date. Search and other tool calls may incur additional charges.
Evaluation: Speed and latency: Artificial Analysis, Gemini 4 Argon (high). Time to the first answer token may include thinking time. Prices come from a separate OpenRouter route, which should not be assumed to match the measurement route.
Artificial Analysis Intelligence Index v5.0.1. Model configuration in the source: Gemini 4 Argon (high). This is an index calculated from multiple tests, not a success percentage. The source was checked on November 12, 2026; this has not been used as the test date.
A further-trained Pro version of Kimi K3. This open-weight multimodal model focuses on long coding and tool-use tasks.
AA intelligence score
46
Context
1,048,576 token
Input / output
$2 / $10 / 1M token
Generation speed
32 tokens/s
Time to first token
179.3 s
Output limit
943,718 token
Published
Not disclosed
License
kimi-k3.1-pro
Input types
Text, Image, Video
Prices are in USD per 1 million tokens for this API route on OpenRouter. Context and output limits apply to this route, not to all services from the developer. The catalog date has not been used as the model release date.
Evaluation: Speed and latency: Artificial Analysis, Kimi K3.1 Pro. Time to the first answer token may include thinking time. Prices come from a separate OpenRouter route, which should not be assumed to match the measurement route.
Artificial Analysis Intelligence Index v5.0.1. Model configuration in the source: Kimi K3.1 Pro. This is an index calculated from multiple tests, not a success percentage. The source was checked on November 12, 2026; this has not been used as the test date.
Publisher's model card: Kimi K3.1 Pro(max) column, GPQA-Diamond row. A developer evaluation, not an independent measurement. Kimi K3.1 Pro max configuration. The test date was not disclosed; the source was checked on November 12, 2026.
Publisher's model card: Kimi K3.1 Pro(max) column, Terminal-Bench 4.0 row. A developer evaluation, not an independent measurement. Kimi K3.1 Pro max configuration. The test date was not disclosed; the source was checked on November 12, 2026.
Publisher's model card: Kimi K3.1 Pro(max) column, HLE-Full row. A developer evaluation, not an independent measurement. Kimi K3.1 Pro max configuration. The test date was not disclosed; the source was checked on November 12, 2026.
Publisher's model card: Kimi K3.1 Pro(max) column, HLE-Full w/ tools row. A developer evaluation, not an independent measurement. Kimi K3.1 Pro max configuration. The test date was not disclosed; the source was checked on November 12, 2026.
Publisher's model card: Kimi K3.1 Pro(max) column, MMMU-Pro row. A developer evaluation, not an independent measurement. Kimi K3.1 Pro max configuration. The test date was not disclosed; the source was checked on November 12, 2026.
Publisher's model card: Kimi K3.1 Pro(max) column, BrowseComp row. A developer evaluation, not an independent measurement. Kimi K3.1 Pro max configuration. The test date was not disclosed; the source was checked on November 12, 2026.
Meta's reasoning model for personal and professional agent tasks. It can generate text from text, image, video and audio inputs.
AA intelligence score
44
Context
1,048,576 token
Input / output
$4 / $20 / 1M token
Generation speed
118 tokens/s
Time to first token
25.19 s
Output limit
943,718 token
Published
Not disclosed
License
Özel kullanım koşulları
Input types
Text, Image, Video, Audio
Prices are in USD per 1 million tokens for this API route on OpenRouter. Context and output limits apply to this route, not to all services from the developer. The catalog date has not been used as the model release date. Search and other tool calls may incur additional charges.
Evaluation: Speed and latency: Artificial Analysis, Muse Wonder (max). Time to the first answer token may include thinking time. Prices come from a separate OpenRouter route, which should not be assumed to match the measurement route.
Artificial Analysis Intelligence Index v5.0.1. Model configuration in the source: Muse Wonder (max). This is an index calculated from multiple tests, not a success percentage. The source was checked on November 12, 2026; this has not been used as the test date.
A Haiku model for high-volume work such as summarization, subagents and browser tasks. A release developed after Haiku 4.5.
AA intelligence score
36
Context
1,000,000 token
Input / output
$0.1 / $0.5 / 1M token
Generation speed
255 tokens/s
Time to first token
185.81 s
Output limit
128,000 token
Published
Not disclosed
License
Özel kullanım koşulları
Input types
Text, Image
Prices are in USD per 1 million tokens for this API route on OpenRouter. Context and output limits apply to this route, not to all services from the developer. The catalog date has not been used as the model release date. Different pricing thresholds apply to long inputs; current thresholds appear in the linked pricing table. Search and other tool calls may incur additional charges.
Evaluation: Speed and latency: Artificial Analysis, Claude Haiku 5.5 (max). Time to the first answer token may include thinking time. Prices come from a separate OpenRouter route, which should not be assumed to match the measurement route.
Artificial Analysis Intelligence Index v5.0.1. Model configuration in the source: Claude Haiku 5.5 (max). This is an index calculated from multiple tests, not a success percentage. The source was checked on November 12, 2026; this has not been used as the test date.
Meta's small Muse model for tool calling, routing and coding subtasks. Designed for repeated agent workloads.
AA intelligence score
36
Context
1,048,576 token
Input / output
$0.1 / $0.5 / 1M token
Generation speed
452 tokens/s
Time to first token
5.05 s
Output limit
943,718 token
Published
Not disclosed
License
Özel kullanım koşulları
Input types
Text, Image, Video, Audio
Prices are in USD per 1 million tokens for this API route on OpenRouter. Context and output limits apply to this route, not to all services from the developer. The catalog date has not been used as the model release date. Search and other tool calls may incur additional charges.
Evaluation: Speed and latency: Artificial Analysis, Muse Ember (max). Time to the first answer token may include thinking time. Prices come from a separate OpenRouter route, which should not be assumed to match the measurement route.
Artificial Analysis Intelligence Index v5.0.1. Model configuration in the source: Muse Ember (max). This is an index calculated from multiple tests, not a success percentage. The source was checked on November 12, 2026; this has not been used as the test date.
The default ranking sorts by the Artificial Analysis Intelligence Index, highest first. The crown marks the highest AA score in the same evaluation profile within this catalog; tied models share the lead. The source specifies each model's reasoning settings.
The AA Intelligence Index combines multiple tests into an index score; it is not a success percentage. Current rankings and methodology: Artificial Analysis. Other benchmarks can be selected independently of the comparison metric. The legacy four-test average is calculated only when GPQA Diamond, MMLU-Pro, SWE-bench Verified and AIME 2025 are all available; it is kept separate from the AA score.
Missing values are not treated as zero, and models without a disclosed AA score cannot lead the intelligence ranking. Prices are USD per million tokens, speed is tokens per second and latency is seconds. Time to first token may include reasoning time. Providers, tools and test budgets can affect results; sources and check dates appear in each model's technical details. The published catalog refreshes every 60 seconds while open.
