Private AI

Nano GPT logo
NanoGPT
Nano GPT logo
NanoGPT
Nano GPT logo
NanoGPT
Search
 
 
 
 
All conversations

Create

Chat
Studio

Explore

Models
Pricing
AI Apps
Blog

Library

Gallery
Workspace
Usage & billing
API
Suggestions and bugs
Use cases

|

Terms

|

Privacy

|

Refunds
Usage & billing
Teams
Referrals
Presets
Settings
FAQPress & mediaBrand assetsBenchmarksNano RPCEarn crypto
  1. Models
  2. Text
  3. Llm360

K2-Think

K2-Think is a 32B open-weights general reasoning model with strong competitive math performance. Benchmarks: AIME 2024 90.83, AIME 2025 81.24, GPQA-Diamond 71.08, LiveCodeBench v5 63.97.

Added Jul 26, 2025

Pricing

Auto routing · per 1M tokens
Input
$0.17
Output
$0.68
Cache read
$0.085
Compare provider prices

Specifications

Context window
128K
Max output
32.8K
Parameters
32B

Benchmarks

No public benchmark scores for this model yet.

Providers

Auto routing is available for this model. Explicit provider selection is not available.

Auto
Auto
Recommended
$0.17$0.68$0.085N/AN/A
Batch API
ProviderInputOutputCached input
Auto
AutoLLM360/K2-Think
$0.17/M$0.68/M$0.085/M

Batch rows for this model use normal routing and cost the same as standard API requests. Set provider in a row to choose a provider.

Reliability

Related text models

Compare K2-Think with similar models from the same provider or model family.

ERNIE Light DT 1.0 (Intl)

baidu/ernie-light-dt-1.0-intl

ERNIE Light DT 1.0 (Intl) is Baidu's text model with Baidu Sparse Attention for working with long conversations and documents. It supports tool calling, structured output, a 1,024,000-token context window, and up to 131,072 output tokens. Choose max, high, or low reasoning effort, or none to disable reasoning.

Gemini 4 Carbon

google/gemini-4-carbon

Google's Gemini 4 Carbon is a reasoning model for complex coding, scientific and mathematical problem solving, and multi-step tasks that use tools. It accepts text, images, audio, video, and PDFs and produces text, with tool calling and structured outputs. Its 1,048,576-token context is shared across input, thinking, and output. Responses support up to 32,768 combined thinking and output tokens, with low, medium, and high thinking levels; medium is the default.

Muse Ember

meta/muse-ember

Meta's Muse Ember is the smallest and most efficient Muse model, built for the work agents repeat all day: tool calls, coding-agent subtasks, routing, and extraction. It accepts text, images, audio, video, and PDF documents across a one-million-token context window, supports tool calling and structured output, and always reasons before answering.

Claude Fable 5.2

anthropic/claude-fable-5.2

Anthropic's Claude Fable 5.2 supports demanding reasoning, coding, and long-running agentic work, with text, image, and PDF inputs, tool calling, and structured outputs. It has a 1M-token context window and up to 128K output tokens. Adaptive thinking is always on, with five effort levels. Prompts and outputs are retained for 30 days; Zero Data Retention is not available.

Nex N2.6

nex-agi/nex-n2.6

Nex N2.6 is Nex AGI's agentic model for coding and long-running tasks that use tools. It accepts text and images, produces text, and supports tool calling and structured output within a 262,144-token context window. Reasoning is always enabled, with medium and high effort levels.

OrcaCyber DeepSeek V4.1 Flash

orca/orcacyber-flash-1.0

An OrcaCyber model built on DeepSeek V4.1 Flash for authorized security research, coding, and tool use. Accepts text and images, with reasoning off by default. Supports a 1M-token context window, up to 393K output tokens, tool calling, and structured output. Offensive work is permitted only within an authorized security engagement.