搜尋模型

Anthropic's most capable generally available Claude model for ambitious, long-running knowledge work and coding projects

Anthropic's best balance of speed and intelligence, reaching near-Opus quality on coding and agentic work

Anthropic's most capable model for complex reasoning, long-horizon agentic coding, and high-autonomy professional work

Anthropic's fastest model with enhanced capabilities and efficiency

Anthropic's most capable model with breakthrough performance across all tasks

Anthropic's most capable commercial model with frontier coding, vision, and long-horizon agentic performance

Anthropic's gated frontier research preview with step-change autonomous cybersecurity capabilities (Project Glasswing access only)

Anthropic's most advanced model with adaptive thinking and frontier-level reasoning across all domains

Anthropic's default model for Claude.ai with excellent balance of capability and speed

Anthropic's most powerful frontier model with exceptional reasoning, coding, and scientific capabilities

Anthropic's most advanced model with enhanced capabilities

Anthropic's next-generation balanced model with superior coding and reasoning

World's best coding model with multi-hour agentic workflows

Anthropic's fastest and most cost-effective Claude 3.5 model

Anthropic's enhanced Claude 3.5 with improved reasoning and coding

Anthropic's most capable model with exceptional reasoning abilities

Anthropic's balanced model offering strong performance and speed

Anthropic's fastest model optimized for speed and efficiency

Anthropic's previous generation fast model

Previous generation Claude with strong safety focus

DeepSeek's frontier MoE flagship closing the gap with leading proprietary models on reasoning and agentic coding

DeepSeek's smaller, fast variant of V4 — same architecture at a fraction of the cost and latency

DeepSeek's latest flagship model matching GPT-5 performance with integrated tool-use thinking

DeepSeek's competition-focused variant (EXPIRED Dec 15, 2025 - was temporary API-only release)

DeepSeek's hybrid model combining V3 and R1 strengths

DeepSeek's upgraded reasoning model with 87.5% AIME accuracy and significantly reduced hallucinations

DeepSeek's updated V3 model with improved post-training using R1 techniques

DeepSeek's reasoning model competing with OpenAI o1, released under MIT license

DeepSeek's latest flagship model with enhanced capabilities and efficiency

DeepSeek's specialized coding model with advanced programming capabilities

DeepSeek's flagship MoE model with exceptional efficiency

DeepSeek's mathematics-specialized model

Google's frontier Flash model built for agentic workflows, coding, and long-horizon tasks at high speed

Google's most cost-efficient Gemini 3 model with 1M context, multimodal input, and 2.5x faster time-to-first-token than Gemini 2.5 Flash

Google's newest flagship with doubled reasoning performance and 1M token context for agentic workflows

Google's enhanced reasoning mode with extended thinking capabilities for complex problems

Google's frontier intelligence model built for speed - now the default model in Gemini app

Google's latest flagship model with advanced multimodal capabilities and PhD-level reasoning

Google's most efficient workhorse model designed for speed with thinking capabilities

Google's most intelligent AI model with advanced thinking capabilities

Google's experimental 2.0 Pro model with advanced capabilities (became GA in February 2025)

Google's latest generation multimodal model with enhanced capabilities and native tool use

Google's fast and efficient multimodal model

Google's flagship multimodal AI model with ultra-long context

Google's previous generation flagship model

Google's previous generation large language model

Meta Superintelligence Labs' strongest model for agentic and coding work — the first Muse model developers can build on, via the new Meta Model API

Meta Superintelligence Labs' first model — a natively multimodal reasoning model with contemplating mode and multi-agent orchestration

Meta's flagship multimodal model (RESEARCH PREVIEW ONLY - weights not publicly released)

Meta's balanced multimodal MoE model with 128 experts for general use

Meta's efficient multimodal model with industry-leading 10M token context

Meta's latest improved Llama model with enhanced capabilities

Meta's largest and most capable open-source model

Meta's updated flagship open-source model

Meta's efficient smaller model with extended context

Meta's original flagship open-source model

Meta's specialized code generation model

Meta's previous generation flagship open-source model

MiniMax's natively multimodal open-weight flagship uniting frontier reasoning and agentic capability in one model, with sparse attention for ultra-long context

MiniMax's self-evolving agent model pioneering recursive self-improvement with frontier agentic coding performance at a fraction of competitor cost

MiniMax's flagship model matching frontier performance at 1/20th the cost with 80.2% SWE-bench Verified

Ultra-fast variant of M2.5 generating 100 tokens per second at $1/hour continuous operation

Lightweight coding-focused model with strong multi-language programming and agentic capabilities

MiniMax's general-purpose foundation model with strong tool use and deep search capabilities

World's first open-source large-scale hybrid-attention reasoning model with 1M token context

Mistral's 128B dense merged model unifying instruction-following, reasoning, and coding with configurable reasoning effort

Mistral's unified open-weight MoE model combining reasoning, multimodal, and coding capabilities under Apache 2.0 with only 6.5B active parameters

Mistral's state-of-the-art open-weight frontier model with multimodal and multilingual capabilities under Apache 2.0

Mistral's high-performance dense model in the new Ministral 3 family

Mistral's efficient edge-ready model for drones, cars, robots, phones and laptops

Mistral's ultra-compact model for resource-constrained edge deployments

Mistral's next-generation coding model designed to compete with Anthropic and other coding-focused LLMs

Mistral's compact open-source coding model that runs on consumer hardware under Apache 2.0

Mistral's flagship reasoning model with advanced multi-step logic capabilities

Mistral's open-source reasoning model with Apache 2.0 license

Mistral's ultra-compact model for edge deployment

Mistral's efficient model for laptops and edge devices with enhanced capabilities

Mistral's most advanced flagship model

Mistral's specialized code generation model

Mistral's largest mixture-of-experts model

Mistral's efficient mixture-of-experts model

Mistral's compact and efficient base model

Moonshot's 2.8T open-weight MoE flagship — the largest open model to date, with native vision, always-on thinking, and #1 frontend coding performance

Moonshot's open-weight 1T MoE flagship for long-horizon coding and multi-agent orchestration with 300-subagent Agent Swarm system

Moonshot AI's open-source visual agentic model with native multimodal capabilities and multi-agent orchestration

Moonshot AI's flagship model - #1 open-source LLM on LMSYS Arena with exceptional agentic capabilities

Moonshot AI's advanced reasoning model matching OpenAI o1 performance

Moonshot AI's flagship chatbot with industry-leading long context capabilities

Moonshot AI's original chatbot launch with 200K Chinese character support

OpenAI's flagship GPT-5.6 model — its most capable system yet for complex coding, cybersecurity, and scientific reasoning, with new max reasoning effort and ultra subagent mode

The balanced mid-tier of the GPT-5.6 family — matches GPT-5.5 quality on high-volume business tasks at roughly half the price

The fastest, most cost-efficient GPT-5.6 model — strong everyday capability at OpenAI's lowest price for summarization, drafting, and routine automation

OpenAI's most capable agentic coding model for complex multi-file development tasks

OpenAI's specialized coding model for Codex agent, tuned for long-horizon multi-file work

OpenAI's highest-accuracy variant of GPT-5.5, tuned for complex research, advanced mathematics, and retrieval-heavy workflows

OpenAI's new default frontier model — sharper reasoning, native computer use, and a major leap in long-context retrieval at matched GPT-5.4 latency

OpenAI's most capable frontier model unifying reasoning, coding, and agentic computer use in a single system

Ultra-fast variant of GPT-5.3-Codex optimized for rapid iteration and quick coding tasks

OpenAI's smartest model with breakthrough benchmark performance - first model above 90% on ARC-AGI

OpenAI's unified intelligence model combining advanced reasoning with multimodal capabilities

OpenAI's first open-weight reasoning model for teams wanting to run and customize models on their own infrastructure

OpenAI's efficient open-weight reasoning model for resource-constrained deployments

OpenAI's most powerful reasoning model designed to think longer and provide the most reliable responses

OpenAI's next-generation reasoning model with breakthrough capabilities

OpenAI's most advanced small reasoning model, best-performing on AIME 2024 and 2025

OpenAI's developer-focused model with extended context capabilities and superior coding performance

OpenAI's efficient mid-size model with 1M token context at a fraction of GPT-4.1 cost

OpenAI's first nano-class model optimized for ultra-low-latency classification and autocompletion

OpenAI's cost-efficient reasoning model available on free tier with strong STEM performance

OpenAI's full reasoning model with advanced problem-solving capabilities

OpenAI's advanced reasoning model trained with reinforcement learning

OpenAI's efficient mini reasoning model optimized for STEM tasks

OpenAI's efficient small model with strong performance

OpenAI's omni-modal flagship model with native multimodal capabilities

OpenAI's latest and most efficient GPT-4 variant

OpenAI's breakthrough multimodal large language model

OpenAI's optimized and cost-effective conversational model

Multimodal agent sibling to Qwen3.7-Max, adding native vision while keeping the 1M-token context and agentic tooling at a lower price point

Alibaba's closed text flagship for reasoning and long-horizon agents — 1M context, MCP and multi-agent orchestration, tuned for autonomous coding and office automation

Alibaba's first dense open-weight Qwen3.6 — 27B beats 397B MoE on agentic coding via hybrid Gated DeltaNet + Gated Attention

Alibaba's most powerful Qwen yet — preview flagship topping six major coding benchmarks

Alibaba's flagship agentic AI model with hybrid linear attention, always-on reasoning, and autonomous multi-step coding workflows

Alibaba's hosted flagship combining hybrid linear-attention MoE with native multimodal understanding for agentic workflows across 201 languages

Alibaba's flagship model with over 1 trillion parameters and exceptional reasoning

Efficient MoE model with 30B total but only 3B active parameters

Alibaba's flagship open-source MoE model with 235B total parameters and 22B active

Reasoning-focused model with extended thinking capabilities

Vision-language model with strong multimodal understanding

Alibaba's instruction-tuned flagship open-source model

SpaceXAI's coding and agentic-work flagship on the new V9 foundation — pitched as Opus-class capability but faster, more token-efficient, and lower cost

xAI's reasoning flagship with always-on reasoning, native video input, 1M context, and significantly cheaper API pricing

xAI's flagship multi-agent model combining four specialized agents on a shared MoE backbone for industry-leading speed and lowest hallucination rates

xAI's enhanced model focusing on personality refinement and real-world usability improvements

World's most intelligent model with native tool use and real-time search

xAI's premium variant with parallel-agent orchestration for the most demanding tasks

xAI's advanced model trained with 10x more compute than Grok-2

xAI's most advanced model with real-time information access

xAI's efficient smaller model with real-time capabilities

Enhanced version of Grok with improved reasoning capabilities

xAI's debut model with unique personality and real-time access

Thinking Machines Lab's first open model — a multimodal Mixture-of-Experts transformer with controllable reasoning effort

Compact variant of Inkling for cost-conscious deployments, retaining the multimodal MoE architecture

Zhipu AI's MIT-licensed flagship extending GLM-5.1 to a 1M-token context, the highest-ranked open-source model on long-horizon coding benchmarks

Zhipu AI's current flagship refined from GLM-5 for agentic engineering, topping the SWE-Bench Pro leaderboard with sustained 8-hour autonomous execution

Zhipu AI's open-weight 744B MoE foundation model purpose-built for agentic engineering, trained entirely on Huawei Ascend chips

Z.ai's flagship model with industry-leading coding and multi-step task handling

Open-source vision-language model optimized for multimodal reasoning and frontend automation

Language model built entirely on Chinese domestic chips

Vision-language model compatible with Huawei Ascend processors

Z.ai's general-purpose flagship trained on 22 trillion tokens