AICraft Model Catalogue · 100+ Models Aggregated · One API Key

AICraft Model Catalogue · Complete Model List (AI-readable version)

AICraft's aggregated API gives you access to 100+ models from 29 providers through a single endpoint — including DeepSeek, Qwen, GLM, Claude, Gemini, MiniMax and Doubao. The complete list follows:

• MiniMax-M1 (MiniMax · text · 1.0M context · input $0.6406/M tokens · output $2.5623/M tokens. Open-source hybrid reasoning model: 1M-token input, up to 80K reasoning output.)
• Claude-4.7-Opus (Anthropic · text · 1.0M context · input $5.5250/M tokens · output $27.6250/M tokens. Built for long-running async agents: large codebases, multi-stage debugging, project orchestration.)
• Claude-4.8-Opus (Anthropic · text · 1.0M context · input $5.5250/M tokens · output $27.6250/M tokens. Opus model with 1M context for multi-step reasoning, complex coding and long tasks.)
• Claude-Fable-5 (Anthropic · text · 1.0M context · input $11.3000/M tokens · output $56.5000/M tokens. Coding and knowledge-work model with 1M context. Data retained 30 days for abuse checks.)
• Claude-Fable-5.1 (Anthropic · text · 1.0M context · input $11.0500/M tokens · output $55.2500/M tokens. Fable 5.1: 1M context, extended thinking and long-horizon coding, research and knowledge work.)
• Claude-Haiku-5.5 (Anthropic · text · 1.0M context · input $0.1105/M tokens · output $0.5525/M tokens. Fast, low-cost Claude 5 generation: 1M context with near-frontier reasoning and coding at high throughput.)
• Claude-Opus-5 (Anthropic · text · 1.0M context · input $5.6500/M tokens · output $28.2500/M tokens. Flagship for hard reasoning, complex coding and long-horizon agents. Orchestrates parallel sub-agents.)
• Claude-Opus-5.5 (Anthropic · text · 1.0M context · input $4.4200/M tokens · output $22.1000/M tokens. Opus 5.5: 1M context, deep adaptive reasoning for complex software engineering and long autonomous runs.)
• Claude-Sonnet-5 (Anthropic · text · 1.0M context · input $2.2600/M tokens · output $11.3000/M tokens. Adaptive thinking (low to x-high effort), 1M context, and text, image and file input.)
• Claude-Sonnet-5.5 (Anthropic · text · 1.0M context · input $2.2100/M tokens · output $11.0500/M tokens. Balanced Claude 5 generation: 1M context, strong coding and agent work at mid-tier pricing.)
• Doubao-Seed-2.1-pro (ByteDance · text · 256K context · input $0.9609/M tokens · output $4.8044/M tokens. Production-grade model: upgraded coding, agent and multimodal abilities, with stronger planning.)
• Doubao-Seed-2.1-turbo (ByteDance · text · 256K context · input $0.4804/M tokens · output $2.4022/M tokens. Cost and performance balanced, with upgraded coding, agent and multimodal abilities.)
• Doubao-Seedance-2.0 (ByteDance · video · input -/M tokens · output $8.1674/M tokens. Video model taking image, video and audio as reference inputs. Also edits and extends video.)
• Doubao-seedance-2-0-fast-260128 (ByteDance · video · input -/M tokens · output $5.9254/M tokens. Faster, cheaper Seedance 2.0: image, video and audio references, editing and extension. Up to 720p.)
• Doubao-seedance-2-0-mini-260615 (ByteDance · video · input -/M tokens · output $3.6833/M tokens. Compact Seedance 2.0 video model: image, video and audio references with editing and extension.)
• Dreamina-seedance-2-0-260128 (ByteDance · video · input -/M tokens · output $8.5085/M tokens. Dreamina Seedance 2.0: multimodal video generation and editing from image, video and audio inputs.)
• Dreamina-seedance-2-0-fast-260128 (ByteDance · video · input -/M tokens · output $6.1880/M tokens. Faster Dreamina Seedance 2.0: lower latency video generation and editing with the same reference inputs.)
• Dreamina-seedance-2-0-mini-260615 (ByteDance · video · input -/M tokens · output $3.8675/M tokens. Small Dreamina Seedance 2.0: budget video generation and editing from image, video and audio references.)
• Claude-4.0-Opus (Anthropic · text · 200K context · input $16.5750/M tokens · output $82.8750/M tokens. Claude 4.0 Opus: 200K context, strong reasoning and coding for demanding analytical and engineering tasks.)
• Claude-4.0-Sonnet (Anthropic · text · 200K context · input $3.3150/M tokens · output $16.5750/M tokens. Claude 4.0 Sonnet: 200K context, balanced cost for general coding, analysis and writing.)
• Claude-4.1-Opus (Anthropic · text · 200K context · input $16.5750/M tokens · output $82.8750/M tokens. Claude 4.1 Opus: 200K context, refined 4.x reasoning with stronger code understanding.)
• Claude-4.5-Haiku (Anthropic · text · 200K context · input $1.1050/M tokens · output $5.5250/M tokens. Fast, low-cost model: near-frontier reasoning and coding at much lower latency and cost.)
• Claude-4.5-Opus (Anthropic · text · 200K context · input $5.5250/M tokens · output $27.6250/M tokens. Frontier reasoning model for complex software engineering and agent workflows. Adjustable effort.)
• Claude-4.5-Sonnet (Anthropic · text · 200K context · input $3.3150/M tokens · output $16.5750/M tokens. Agent and coding model: system design, code security and long autonomous runs.)
• Claude-4.6-Opus (Anthropic · text · 1.0M context · input $5.5250/M tokens · output $27.6250/M tokens. Opus model with 128K max output and adaptive thinking depth. Supports a max-effort mode.)
• Claude-4.6-Sonnet (Anthropic · text · 1.0M context · input $3.3150/M tokens · output $16.5750/M tokens. Coding and agent model: iterative development, codebase navigation and reliable computer use.)
• DeepSeek-V4-Flash-20260731 (DeepSeek · text · 1.0M context · input $0.3203/M tokens · output $1.2812/M tokens. V4 Flash checkpoint with stronger agent ability, native Responses API and Codex support.)
• DeepSeek-V4-Flash-Vision-Exp (DeepSeek · text · 1.0M context · input $0.4804/M tokens · output $1.4413/M tokens. Experimental vision-language model released August 2026.)
• DeepSeek-V4-Pro-0813 (DeepSeek · text · 1.0M context · input $1.4413/M tokens · output $4.3239/M tokens. Flagship MoE model with a 1M context. Built for high-complexity agent and coding tasks.)
• DeepSeek-V4.1-Flash (DeepSeek · text · 1.0M context · input $0.3203/M tokens · output $1.2812/M tokens. Updated V4.1 Flash: 1M context and higher throughput for high-volume generation.)
• Doubao-1.5-Pro-32k (ByteDance · text · 128K context · input $0.1281/M tokens · output $0.3203/M tokens. Multimodal model: visual understanding, classification and information extraction. 128K context.)
• Doubao-1.5-Thinking-Pro (ByteDance · text · 128K context · input $0.6406/M tokens · output $2.5623/M tokens. Doubao reasoning model: long chains of thought for math, logic and multi-step analysis. 128K context.)
• Doubao-1.5-Vision-Pro (ByteDance · text · 128K context · input $0.4804/M tokens · output $1.4413/M tokens. Doubao vision-language model: reads images, charts and documents alongside text. 128K context.)
• Doubao-Seed-1.6 (ByteDance · text · 256K context · input $0.1281/M tokens · output $0.3203/M tokens. Doubao Seed 1.6: 256K context general model with solid Chinese and English understanding.)
• Doubao-Seed-1.6-Flash (ByteDance · text · 256K context · input $0.0240/M tokens · output $0.2402/M tokens. Cheapest Seed 1.6 tier: 256K context, fast responses for bulk and real-time workloads.)
• Doubao-Seed-1.6-Thinking (ByteDance · text · 256K context · input $0.1281/M tokens · output $1.2812/M tokens. Seed 1.6 reasoning mode: visible chain of thought for multi-step problems. 256K context.)
• Doubao-Seed-2.0-Code (ByteDance · text · 256K context · input $0.5125/M tokens · output $2.5623/M tokens. Coding model for real IDE environments with tool calling, tuned for frontend work. Supports Skills.)
• Doubao-Seed-2.0-Lite (ByteDance · text · 256K context · input $0.0961/M tokens · output $0.5765/M tokens. Balanced cost and performance: unstructured data, content creation, search and data analysis.)
• Doubao-Seed-2.0-Mini (ByteDance · text · 256K context · input $0.0320/M tokens · output $0.3203/M tokens. Low-latency model for high-concurrency work. 256K context, 4 thinking-length levels.)
• Doubao-Seed-2.0-Pro (ByteDance · text · 256K context · input $0.5125/M tokens · output $2.5623/M tokens. Flagship general model for agent-era reasoning: multi-step planning, video and image-text reasoning.)
• Gemini-2.5-Flash (Google · text · 1.0M context · input $0.3315/M tokens · output $2.7625/M tokens. Visual generation model for creative design and multimodal understanding.)
• Gemini-2.5-Flash-Image (Google · image · 32K context · input $0.3315/M tokens · output $33.1500/M tokens. Image generation and editing model (aka Nano Banana) with multi-turn conversational editing.)
• Gemini-2.5-Flash-Lite (Google · text · 1.0M context · input $0.1105/M tokens · output $0.4420/M tokens. Low-latency reasoning model for high-throughput, cost-sensitive use. Multi-turn thinking off by default.)
• Gemini-2.5-Pro (Google · text · 1.0M context · input $1.3813/M tokens · output $11.0500/M tokens. Reasoning model for coding, math and science tasks, with a 1M-token context window.)
• Gemini-3.0-Flash-Preview (Google · text · 1.0M context · input $0.5525/M tokens · output $3.3150/M tokens. Preview build of Gemini 3.0 Flash.)
• Gemini-3.0-Pro-Image-Preview (Google · image · 32K context · input $2.2100/M tokens · output $132.6000/M tokens. Preview build of Gemini 3.0 Pro for image generation: 32K context, token-based pricing.)
• Gemini-3.1-Flash-Image-Preview (Google · image · 131K context · input $0.5525/M tokens · output $66.3000/M tokens. Preview Gemini 3.1 Flash image model: 131K context, token-based pricing at the cheaper Flash tier.)
• Gemini-3.1-Flash-Lite-Preview (Google · text · 1.0M context · input $0.2762/M tokens · output $1.6575/M tokens. Lightweight preview of Gemini 3.1: 1M context at low cost for classification and short generation.)
• Gemini-3.1-Pro-Preview (Google · text · 1.0M context · input $2.2100/M tokens · output $13.2600/M tokens. Reasoning model for complex workflows: software engineering, agent reliability, multimodal input.)
• GLM-4.5 (zAI · text · 131K context · input $0.6406/M tokens · output $2.5623/M tokens. GLM 4.5: hybrid reasoning model with 131K context for coding and agent workflows.)
• Gemini-3.0-Pro-Image (Google · image · 32K context · input $2.2100/M tokens · output $132.6000/M tokens. Image model for complex, multi-turn generation and editing, with reasoning before each output.)
• Gemini-3.1-Flash-Image (Google · image · 131K context · input $0.5525/M tokens · output $66.3000/M tokens. Image generation and editing model (aka Nano Banana 2) at Flash speed and cost.)
• Gemini-3.1-Flash-Lite (Google · text · 1.0M context · input $0.2762/M tokens · output $1.6575/M tokens. Efficient model for high traffic: audio input, RAG reranking, translation, code completion. All thinking levels.)
• Gemini-3.1-Flash-Lite-Image (Google · image · 66K context · input $0.2762/M tokens · output $33.1500/M tokens. Fastest, cheapest Google image model (~4s text-to-image). Editing, composition, 1K output.)
• Gemini-3.5-Flash (Google · text · 1.0M context · input $1.6575/M tokens · output $9.9450/M tokens. Fast model with agentic ability close to pro level. Reasons before it answers.)
• Gemini-3.5-Flash-Lite (Google · text · 1.0M context · input $0.3315/M tokens · output $2.7625/M tokens. Efficient model suited as a sub-agent for targeted tasks in multi-agent workflows.)
• Gemini-3.6-Flash (Google · text · 1.0M context · input $0.8287/M tokens · output $4.1437/M tokens. Model for coding, agent workflows, web and app development. Good results with fewer tokens and calls.)
• Gemini-3.7-Flash (Google · text · 1.0M context · input $0.8287/M tokens · output $4.1437/M tokens. Better debugging and first-pass code accuracy than Gemini 3.6 Flash, at half the initial price.)
• Gemini-3.8-Flash (Google · text · 1.0M context · input $0.8287/M tokens · output $4.1437/M tokens. Gemini 3.8 Flash: 1M context, fast multimodal reasoning for everyday production traffic.)
• Gemini-Nano-Banana-2.1 (Google · image · 131K context · input $1.6575/M tokens · output $33.1500/M tokens. Gemini Nano Banana 2.1 image model: 131K context, token-based pricing on the smaller Google tier.)
• gpt-oss-120b (OpenAI · text · 128K context · input $0.1657/M tokens · output $0.8287/M tokens. Open-weight 117B MoE (5.1B active) that runs on one H100. Configurable reasoning, native tools.)
• Kimi-K2 (Moonshot-Kimi · text · 128K context · input $0.6406/M tokens · output $2.5623/M tokens. Kimi K2: open-weight MoE model tuned for agentic tool use and long documents. 128K context.)
• Kling-Image-O1 (Kuaishou · image · input -/M tokens · output -/M tokens. Image editing with strong reference control: changes subject, style and detail, keeps consistency.)
• Kling-v2-1 (Kuaishou · video · 4K context · input -/M tokens · output -/M tokens. Image model with 8 aspect ratios, from 1:1 and 16:9 through to 9:16 and 21:9.)
• Kling-V2-5-Turbo (Kuaishou · video · 10.0M context · input -/M tokens · output -/M tokens. Video model with better causal and temporal instruction following, smoother motion, stable camera.)
• Kling-V2-6 (Kuaishou · video · 100.0M context · input -/M tokens · output -/M tokens. Video model with native audio-video sync and motion control captured from a reference video.)
• Kling-V3 (Kuaishou · video · input -/M tokens · output -/M tokens. Video model producing up to 15 seconds of HD footage with multi-shot coherence and native audio.)
• Kling-Video-O1 (Kuaishou · video · 9K context · input -/M tokens · output -/M tokens. Unified video model for generation, editing and understanding, driven by simple conversation.)
• Microsoft Decision 1 (Microsoft · text · 33K context · input $0.0464/M tokens · output -/M tokens. Microsoft Decision 1: 33K context decision model, input-only token pricing.)
• MiniMax-M2 (Minimax · text · 200K context · input $0.3363/M tokens · output $1.3452/M tokens. MiniMax M2: 200K context, balanced reasoning and writing with strong Chinese support.)
• MiniMax-M2.1 (Minimax · text · 204K context · input $0.3363/M tokens · output $1.3452/M tokens. MiniMax M2.1: 204K context refinement of M2 with steadier long-form output.)
• MiniMax-M2.5 (MiniMax · text · 204K context · input $0.3363/M tokens · output $1.3452/M tokens. Built for agent work, with strong Office abilities: Excel processing, slide generation and research.)
• MiniMax-M2.5-Highspeed (MiniMax · text · 204K context · input $0.6726/M tokens · output $2.6904/M tokens. High-throughput version of MiniMax M2.5.)
• MiniMax-M2.7 (MiniMax · text · 204K context · input $0.3363/M tokens · output $1.3452/M tokens. Model for autonomous work: plans, executes and refines complex tasks in dynamic environments.)
• MiniMax-M3 (MiniMax · text · 1.0M context · input $0.3363/M tokens · output $1.3452/M tokens. Multimodal model (text, image, video in) with 1M context. Sparse attention cuts long-context cost.)
• Kimi-K2-Thinking (Moonshot-Kimi · text · 256K context · input $0.6406/M tokens · output $2.5623/M tokens. Kimi K2 in thinking mode: 256K context and a visible reasoning trace for complex work.)
• Kimi-K2.5 (Moonshot · text · 256K context · input $0.6406/M tokens · output $3.3630/M tokens. Multimodal model with vision and text input, plus thinking and non-thinking modes for chat and agents.)
• Kimi-K2.6 (Moonshot · text · 262K context · input $1.0409/M tokens · output $4.3239/M tokens. Multimodal model with text, image and video input. Strong at agent and coding work.)
• Kimi-K2.7-Code (Moonshot · text · 262K context · input $1.0409/M tokens · output $4.3239/M tokens. Coding MoE (1T total, 32B active) with text and image input. Always-on thinking, 256K context.)
• Kimi-K3 (Moonshot · text · 1.0M context · input $3.2029/M tokens · output $16.0145/M tokens. Open-source 3T-class flagship with native vision and 1M context for long-horizon coding.)
• GPT-5 (OpenAI · text · 400K context · input $1.3813/M tokens · output $11.0500/M tokens. Reasoning model with test-time routing and reduced hallucination. Strong at coding and writing.)
• GPT-5-Chat (OpenAI · text · 128K context · input $1.3813/M tokens · output $11.0500/M tokens. GPT-5 tuned for conversation: natural dialogue and reliable instruction following. 128K context.)
• GPT-5-Mini (OpenAI · text · 400K context · input $0.2762/M tokens · output $2.2100/M tokens. Compact GPT-5 for lightweight reasoning: same instruction following at lower latency and cost.)
• GPT-5-Nano (OpenAI · text · 400K context · input $0.0553/M tokens · output $0.4420/M tokens. Smallest, fastest GPT-5 variant for low-latency tools and real-time use, at the lowest cost.)
• GPT-5-Pro (OpenAI · text · 400K context · input $16.5750/M tokens · output $132.6000/M tokens. High-accuracy GPT-5 for step-by-step reasoning and high-stakes tasks. Supports test-time routing.)
• GPT-5.2 (OpenAI · text · 400K context · input $1.9337/M tokens · output $15.4700/M tokens. Flagship with adaptive reasoning: scales compute to task complexity. Strong at math and code.)
• GPT-5.2-Chat (OpenAI · text · 128K context · input $1.9337/M tokens · output $15.4700/M tokens. Fast, conversational GPT-5.2 for low-latency, high-throughput chat. Reasons only when needed.)
• GPT-5.2-Codex (OpenAI · text · 400K context · input $1.9337/M tokens · output $15.4700/M tokens. Coding model for interactive and long autonomous runs: builds, debugs, refactors and reviews code.)
• GPT-5.3-Codex (OpenAI · text · 400K context · input $1.9337/M tokens · output $15.4700/M tokens. Combines GPT-5.2-Codex coding with GPT-5.2 reasoning. 25% faster and uses fewer tokens.)
• GPT-5.4 (OpenAI · text · 1.1M context · input $2.7625/M tokens · output $16.5750/M tokens. Unified Codex and GPT model with 922K input and 128K output. Strong at coding and document work.)
• GPT-5.4-Mini (OpenAI · text · 400K context · input $0.8287/M tokens · output $4.9725/M tokens. Faster, cheaper GPT-5.4 for high-throughput use: reasoning, coding and tool use at scale.)
• GPT-5.4-Nano (OpenAI · text · 400K context · input $0.2210/M tokens · output $1.3813/M tokens. Lightest GPT-5.4 for low-latency, high-volume work: classification, extraction, ranking, sub-agents.)
• GPT-5.4-Pro (OpenAI · text · 1.0M context · input $33.1500/M tokens · output $198.9000/M tokens. GPT-5.4 Pro adds deeper reasoning for high-stakes work: agentic coding and long-context tasks.)
• GPT-5.5 (OpenAI · text · 1.0M context · input $5.5250/M tokens · output $33.1500/M tokens. Frontier model for complex professional work, with stronger reasoning than GPT-5.4.)
• GPT-5.6-Luna (OpenAI · text · 1.1M context · input $0.2210/M tokens · output $1.3260/M tokens. Smallest, fastest and cheapest model in the GPT-5.6 family.)
• GPT-5.6-Sol (OpenAI · text · 1.1M context · input $5.5250/M tokens · output $33.1500/M tokens. GPT-5.6 flagship for hard reasoning, coding, research, security and long-running agents.)
• GPT-5.6-Terra (OpenAI · text · 1.1M context · input $2.2100/M tokens · output $13.2600/M tokens. Balanced GPT-5.6 model trading some intelligence for lower cost, for everyday production.)
• GPT-6-Astra (OpenAI · text · 1.1M context · input $11.0500/M tokens · output $55.2500/M tokens. Top GPT-6 tier: 1.1M context and the deepest reasoning for research and large-codebase work.)
• GPT-6-Luna (OpenAI · text · 1.1M context · input $0.1105/M tokens · output $0.5525/M tokens. Low-cost GPT-6 tier: 1.1M context at the cheapest GPT-6 rate for high-volume tasks.)
• GPT-6-Sol (OpenAI · text · 1.1M context · input $2.2100/M tokens · output $11.0500/M tokens. Balanced GPT-6 tier: 1.1M context and strong general reasoning at mid pricing.)
• GPT-6.1-Sol (OpenAI · text · 1.1M context · input $2.2100/M tokens · output $11.0500/M tokens. GPT-6.1 Sol: update to the balanced tier with better instruction adherence and consistency.)
• GPT-6.1-Terra (OpenAI · text · 1.1M context · input $1.1050/M tokens · output $5.5250/M tokens. GPT-6.1 Terra: 1.1M context, priced between the Luna and Sol tiers.)
• GPT-Image-2 (OpenAI · image · input $5.5250/M tokens · output $33.1500/M tokens. Image model with multilingual text rendering, stronger instruction following and higher resolution.)
• Qwen2.5-Max-2025-01-25 (Aliyun · text · 128K context · input $0.3844/M tokens · output $1.5374/M tokens. Qwen 2.5 Max snapshot (2025-01-25): 128K context, flagship of the previous Qwen Max generation.)
• Qwen-Turbo (Aliyun · text · 1.0M context · input $0.0480/M tokens · output $0.4804/M tokens. Qwen Turbo: 1M context at very low cost for high-throughput chat and extraction.)
• Qwen-VL-MAX-2025-01-25 (Aliyun · text · 128K context · input $0.4804/M tokens · output $1.4413/M tokens. Qwen VL Max snapshot: vision-language flagship for images, documents and charts. 128K context.)
• Qwen3.5-Plus (Alibaba · text · 1.0M context · input $0.1281/M tokens · output $0.7687/M tokens. Vision-language Plus model with a 1M-token context. Handles text and multimodal input.)
• Qwen3.6-27B (Alibaba · text · 262K context · input $0.3690/M tokens · output $3.6897/M tokens. 27B hybrid multimodal model with a 262K-token context window.)
• Qwen3.6-Plus (Alibaba · text · 1.0M context · input $0.3203/M tokens · output $1.9217/M tokens. Better reasoning and agent behavior than Qwen3.5. Strong at agentic coding and frontend development.)
• Qwen3.7 Flash (Aliyun · text · 1.0M context · input $0.0320/M tokens · output $0.1281/M tokens. Qwen 3.7 Flash: 1M context, the cheapest tier for simple high-volume requests.)
• Qwen3.7-Max (Alibaba · text · 1.0M context · input $1.9217/M tokens · output $5.7652/M tokens. Flagship agent model (text only for now): coding, office tasks and long-horizon autonomous execution.)
• Qwen3.7-Plus (Alibaba · text · 1.0M context · input $0.3203/M tokens · output $1.2812/M tokens. Multimodal agent: reads screens, operates GUIs, writes code from visual references, drives mobile apps.)
• Qwen3.8 Flash (Aliyun · text · 1.0M context · input $0.1601/M tokens · output $0.4804/M tokens. Qwen 3.8 Flash: 1M context, faster and stronger than 3.7 Flash for everyday production.)
• Qwen3.8-Max (Alibaba · text · 1.0M context · input $1.9217/M tokens · output $5.7652/M tokens. 2.4T-parameter MoE flagship: coding, office and other professional work. Runs autonomously for days.)
• Qwen-2.5-VL-72B-Instruct (Aliyun · text · 128K context · input $2.5623/M tokens · output $7.6870/M tokens. Qwen 2.5 VL 72B: dense vision-language model for detailed image and document reading. 128K.)
• Qwen-3-235B-A22B (Aliyun · text · 128K context · input $0.3203/M tokens · output $3.2029/M tokens. Qwen3 235B (22B active): MoE flagship for reasoning, coding and long-context analysis. 128K.)
• Qwen3-235b-A22B-Instruct-2507 (Aliyun · text · 262K context · input $0.3203/M tokens · output $1.2812/M tokens. July 2025 non-thinking Qwen3 235B: direct answers without a reasoning trace. 262K context.)
• Qwen3-235B-A22B-Thinking-2507 (Aliyun · text · 262K context · input $0.3203/M tokens · output $3.2029/M tokens. July 2025 thinking Qwen3 235B: full reasoning trace for hard math, science and code. 262K.)
• Qwen3-30B-A3B (Aliyun · text · 40K context · input $0.1201/M tokens · output $1.2011/M tokens. Qwen3 30B (3B active): small MoE model with optional reasoning modes. 40K context.)
• Qwen3-30b-A3b-Instruct-2507 (Aliyun · text · 128K context · input $0.1201/M tokens · output $0.4804/M tokens. July 2025 non-thinking Qwen3 30B: fast direct responses. 128K context.)
• Qwen3-30b-A3b-Thinking-2507 (Aliyun · text · 126K context · input $0.1201/M tokens · output $1.2011/M tokens. July 2025 thinking Qwen3 30B: compact reasoning model with a visible trace. 126K context.)
• Qwen3-32B (Aliyun · text · 40K context · input $0.3203/M tokens · output $3.2029/M tokens. Qwen3 32B dense model: reliable instruction following and steady reasoning. 40K context.)
• Qwen3-Coder-480B-A35B-Instruct (Aliyun · text · 262K context · input $0.9609/M tokens · output $3.8435/M tokens. Qwen3 Coder 480B: built for repository-scale code generation, refactoring and repair. 262K.)
• Qwen3-Max (Aliyun · text · 262K context · input $0.9609/M tokens · output $3.8435/M tokens. Qwen3 Max: Alibaba's flagship general model with 262K context and the strongest Qwen reasoning.)
• Qwen3-Max-Preview (Aliyun · text · 256K context · input $0.9609/M tokens · output $3.8435/M tokens. Preview build of Qwen3 Max: 256K context and the newest tuning before general release.)
• Qwen3-Next-80B-A3B-Instruct (Aliyun · text · 131K context · input $0.1601/M tokens · output $0.6406/M tokens. Qwen3 Next 80B: next-generation MoE architecture for efficient long-context handling. 131K.)
• Qwen3-Next-80B-A3B-Thinking (Alibaba · text · 131K context · input $0.1601/M tokens · output $1.6015/M tokens. Reasoning-first model: math proofs, code, logic, agent planning. Steadier long chains than earlier Qwen3.)
• Qwen3-Vl-30b-A3b-Thinking (Aliyun · text · 128K context · input $0.1201/M tokens · output $1.2011/M tokens. Qwen3 VL 30B in thinking mode: image understanding with an explicit reasoning trace. 128K context.)
• Qwen3.5-397B-A17B (Alibaba · text · 256K context · input $0.1922/M tokens · output $1.1530/M tokens. Native vision-language model: text, reasoning, code, agents, image, video and GUI.)
• Step-3.7-Flash (StepFun · text · 256K context · input $0.2162/M tokens · output $1.2972/M tokens. Flagship MoE with 256K context, native multimodal understanding and agent orchestration.)
• Hy3-Preview (Tencent · text · 262K context · input $0.1922/M tokens · output $0.6406/M tokens. Efficient MoE for agent workflows with three reasoning levels and strong multi-step code generation.)
• Hy4 Preview (Tencent · text · 1.0M context · input $0.9609/M tokens · output $2.8826/M tokens. Tencent Hy4 preview: 1M context for Chinese-first reasoning, writing and generation.)
• Jev 1.13 (Typesafe · text · 64K context · input $0.0464/M tokens · output -/M tokens. Typesafe Jev 1.13: 64K context, input-token pricing.)
• Jev 1.14 (Typesafe · text · 262K context · input $0.0464/M tokens · output -/M tokens. Typesafe Jev 1.14: 262K context, input-token pricing.)
• Veo-3.1-Fast-Generate-001 (Google · video · input -/M tokens · output -/M tokens. Google video generation model with stable, realistic output at faster speed.)
• Veo-3.1-Generate-001 (Google · video · input -/M tokens · output -/M tokens. Google video generation model with stable, realistic output.)
• Viduq3 (Vidu · video · input -/M tokens · output -/M tokens. Video model generating up to 16-second 1080p clips with audio.)
• Viduq1 (Vidu · video · 99K context · input -/M tokens · output -/M tokens. Fast, low-cost video model for prototyping. Text-to-video and reference-image-to-video.)
• Viduq2 (Vidu · video · 9K context · input -/M tokens · output -/M tokens. Video model balancing speed and quality, 360p to 1080p. Text-to-video and reference-image-to-video.)
• Grok-4-Fast (xAI · text · 2.0M context · input $0.2210/M tokens · output $0.5525/M tokens. Grok 4 Fast: 2M context with adjustable reasoning, fast and cheap access to the large-context line.)
• Grok-4-Fast-Non-Reasoning (xAI · text · 2.0M context · input $0.2210/M tokens · output $0.5525/M tokens. Grok 4 Fast without reasoning: direct responses only, lowest latency in the Grok line. 2M context.)
• Grok-4-Fast-Reasoning (xAI · text · 2.0M context · input $0.2210/M tokens · output $0.5525/M tokens. Grok 4 Fast with reasoning always on: a visible thinking step before the answer. 2M context.)
• Grok-4.1-Fast (xAI · text · 2.0M context · input $0.2210/M tokens · output $0.5525/M tokens. Multimodal model with a 2M-token context window, tuned for low cost per token.)
• Grok-4.1-Fast-Non-Reasoning (xAI · text · 2.0M context · input $0.2210/M tokens · output $0.5525/M tokens. Multimodal model with a 2M-token context window, tuned for low cost per token. Non-reasoning.)
• Grok-4.1-Fast-Reasoning (xAI · text · 20.0M context · input $0.2210/M tokens · output $0.5525/M tokens. Multimodal model with a 2M-token context window, tuned for low cost per token. Reasoning.)
• Grok-4.2-Fast-Non-Reasoning (xAI · text · 2.0M context · input $0.2210/M tokens · output $0.5525/M tokens. Fast flagship with strong agent tool calling and strict instruction following. Non-reasoning.)
• Grok-4.2-Fast-Reasoning (xAI · text · 2.0M context · input $0.2210/M tokens · output $0.5525/M tokens. Fast flagship with strong agent tool calling and strict instruction following. Reasoning.)
• Grok-4.2-Non-Reasoning (xAI · text · input $1.3813/M tokens · output $2.7625/M tokens. Flagship model with strong agent tool calling and strict instruction following.)
• Grok-4.2-Reasoning (xAI · text · 2.0M context · input $2.7625/M tokens · output $0.2210/M tokens. Flagship model with strong agent tool calling and strict instruction following. Reasoning.)
• Grok-4.3 (xAI · text · 1.0M context · input $1.3813/M tokens · output $2.7625/M tokens. Flagship with text and image input and up to 1M context. Built for long documents and automation.)
• Grok-4.5 (xAI · text · 500K context · input $2.2100/M tokens · output $6.6300/M tokens. Strong coding, knowledge-work and STEM performance, with a 500K-token context window.)
• Grok-4.6 (xAI · text · 500K context · input $2.2100/M tokens · output $6.6300/M tokens. Builds on Grok 4.5 with a focus on long-running agents and interactive, visual work.)
• Grok 4.7 (xAI · text · 500K context · input $2.2100/M tokens · output $6.6300/M tokens. Grok 4.7: 500K context flagship for multi-step reasoning, science and current-events analysis.)
• Grok-code-Fast-1 (xAI · text · 256K context · input $0.2210/M tokens · output $1.6575/M tokens. Fast, low-cost reasoning model for agentic coding, with visible reasoning traces you can steer.)
• GLM-4.6 (zAI · text · 200K context · input $1.1530/M tokens · output $2.0178/M tokens. GLM 4.6: 200K context with stronger coding and agent performance than GLM 4.5.)
• GLM-4.7 (zAI · text · 200K context · input $0.5073/M tokens · output $2.0063/M tokens. GLM 4.7: 200K context, the latest GLM generation with improved reasoning and tool use.)
• GLM-5 (Zhipu · text · 200K context · input $0.6406/M tokens · output $2.8826/M tokens. Flagship base model for agentic engineering: complex systems work and long-horizon agent tasks.)
• GLM-5.1 (Zhipu · text · 200K context · input $0.9609/M tokens · output $3.8435/M tokens. Coding model that can run one task for 8+ hours, planning and self-correcting to a finished result.)
• GLM-5.2 (Zhipu · text · 1.0M context · input $1.2812/M tokens · output $4.4841/M tokens. 1M-context flagship for long tasks: stable execution from requirements through to deployment.)
• GLM-5.3 (Zhipu · text · 1.0M context · input $1.2812/M tokens · output $4.4841/M tokens. Coding model with strong security skills: white-box code review and vulnerability discovery.)
• GLM-5.3-Flash (zAI · text · 1.0M context · input $0.1281/M tokens · output $0.4484/M tokens. GLM 5.3 Flash: 1M context, the low-cost fast tier of the GLM 5 line.)