Latest tracked model releases

gemini-3.1-pro-preview

Preview Gemini 3.1 Pro model for advanced intelligence, complex problem solving, agentic workflows, and coding.

gemini-3-flash-preview

Gemini 3 Flash preview model for frontier multimodal performance, agentic tasks, and coding at lower cost.

gemini-2.0-flash-exp

Next generation features, superior speed, native tool use, and multimodal generation

qwen3.6-max-preview

Qwen3.6 frontier preview (sparse MoE, ~1T params), 20% launch discount

gemini-3.1-pro-preview-customtools

Gemini 3.1 Pro Preview endpoint optimized for agentic workflows that use bash and custom tools.

gemini-2.0-flash-thinking-exp

Reasoning model with enhanced thinking capabilities

qwq-32b-preview

Mathematical reasoning specialist with deep thinking capabilities.

gpt-4o-search-preview-2025-03-11

O-search-preview-2025-03-11

gpt-4o-search-preview

O Search PreviewGPT model for web search in Chat Completions

gpt-4o-realtime-preview-2025-06-03

O-realtime-preview-2025-06-03

gpt-4o-realtime-preview-2024-12-17

O-realtime-preview-2024-12-17

gpt-4o-realtime-preview

O RealtimeModel capable of realtime text and audio inputs and outputs

gpt-4o-mini-search-preview-2025-03-11

O-mini-search-preview-2025-03-11

gpt-4o-mini-search-preview

O mini Search PreviewFast, affordable small model for web search

gpt-4o-mini-realtime-preview-2024-12-17

O-mini-realtime-preview-2024-12-17

gpt-4o-mini-realtime-preview

O mini RealtimeSmaller realtime model for text and audio inputs and outputs

gpt-4o-mini-audio-preview-2024-12-17

O-mini-audio-preview-2024-12-17

gpt-4o-mini-audio-preview

O mini AudioSmaller model capable of audio inputs and outputs

gpt-4o-audio-preview-2025-06-03

O-audio-preview-2025-06-03

gpt-4o-audio-preview-2024-12-17

O-audio-preview-2024-12-17

gpt-4o-audio-preview

O AudioGPT-4o models capable of audio inputs and outputs

gemini-3.1-flash-tts-preview

Low-latency Gemini 3.1 speech generation model with natural and steerable audio output.

gemini-3.1-flash-live-preview

Low-latency Gemini 3.1 Live API model for real-time dialogue and voice-first AI applications.

gemini-3.1-flash-lite-preview

Cost-efficient Gemini 3.1 model for high-volume agentic tasks, translation, transcription, and structured extraction.

gemini-2.5-pro-preview-tts

High-fidelity Gemini 2.5 Pro text-to-speech model for structured audio workflows.

gemini-2.5-flash-preview-tts

Fast and controllable Gemini 2.5 text-to-speech model.

gemini-2.5-flash-native-audio-preview-12-2025

Gemini 2.5 Flash Live API model for low-latency voice and video agents with native audio reasoning.

o3-mini

A small model alternative to o3

o3

Reasoning model for complex tasks, succeeded by GPT-5

o1-mini

A small model alternative to o1

o1

Previous full o-series reasoning model

gpt-5.5

GPT model. our newest frontier model for the most complex professional work.

gpt-5.4

GPT model. our frontier model for complex professional work.

gpt-4o-mini

O miniFast, affordable small model for focused tasks

gpt-4o

OFast, intelligent, flexible GPT model

claude-sonnet-4-6

Best combination of speed and intelligence

claude-opus-4-7

Most capable generally available model for complex reasoning and agentic coding

claude-3-5-sonnet-20241022

Most intelligent model, highest level of intelligence and capability

claude-3-5-sonnet-20240620

Structured model profile in ModelMeta.

gemini-2.5-pro

Enhanced mid-generation model with improved capabilities

deepseek-v4-pro

DeepSeek V4 Pro — 1.6T MoE (49B active), 1M context. State-of-the-art open-source model rivaling frontier closed models. Supports function calling, JSON mode, and streaming.

deepseek-v4-flash

DeepSeek V4 Flash — 284B MoE (13B active), 1M context. Replaces deepseek-chat. Open-source, supports function calling, JSON mode, and streaming.

deepseek-r1

DeepSeek's first reasoning model, achieving performance comparable to OpenAI o1 for math, code, and reasoning tasks.

grok-4.1-thinking

xAI's flagship reasoning model, featuring advanced internal monologue and state-of-the-art math/code capabilities.

grok-4.1

xAI's most intelligent general-purpose model, designed for real-time information processing and complex tasks.

llama-3.3-70b-instruct

Meta's most capable 70B model, featuring performance of Llama 3.1 405B at a fraction of the cost.

command-r-plus

Cohere's most powerful model, optimized for RAG and tool use

kimi-k2

Moonshot's next-generation model with superior long-context reasoning and instruction following.