Models

DeepSeek's low-cost V4 API route for 1M-context production assistants, agents, and compatibility migrations.

DeepSeek's stronger V4 route for million-token reasoning, agentic coding, and higher-end open-weight workloads.

Zhipu's GLM long-context model with strong coding ability and open-weight plus API access.

Z.ai's GLM-5.1, a 744B-parameter MoE open-weight model with strong autonomous coding and tool-use behavior.

Z.ai's 1M-context MIT-licensed open-weight flagship for coding and long-horizon agent work.

Original GPT-5 release entry, now superseded in OpenAI guidance by GPT-5.5 and newer GPT-5.4 generation variants.

Older cost-efficient GPT-5 variant now mainly useful for compatibility after GPT-5.4 mini became the recommended fresh-build route.

Older ultra-low-cost GPT-5 tier now mainly useful for compatibility after GPT-5.4 nano became the recommended fresh-build route.

Earlier GPT-5 Codex release entry kept as a historical baseline in OpenAI's Codex model lineage.

Previous GPT-5.2 frontier route now mainly useful for compatibility after GPT-5.5 became OpenAI's recommended fresh-build model.

Previous premium GPT-5.2 reasoning tier now mainly useful for compatibility after GPT-5.5 Pro became OpenAI's newer premium route.

API-published GPT-5.3 Codex route for pinned integrations, deprecated in Codex when using ChatGPT sign-in.

Current GPT-5.4 premium API model for difficult professional work, tool use, and computer-assisted tasks.

OpenAI's strongest mini model for coding, computer use, and fast high-volume agent workloads.

OpenAI's cheapest GPT-5.4 route for fast classification, extraction, and lightweight coding subagents.

Premium GPT-5.4 escalation tier for decision-ready analysis and demanding professional workflows.

Still-available GPT-5.5 API route for pinned agentic coding, professional work, research, and migration baselines.

Still-available premium GPT-5.5 API route for pinned high-compute reasoning and long-running professional workflows.

OpenAI's lowest-cost GPT-5.6 model for high-volume agents, coding support, extraction, and efficient professional workflows.

OpenAI's flagship GPT-5.6 model for complex coding, professional work, science, design, and tool-heavy agents.

Balanced GPT-5.6 model for production coding, knowledge work, agents, and long-context workflows at half Sol's token price.

OpenAI's life-sciences research preview model for biology, drug discovery, and tool-heavy scientific workflows.

Moonshot's Kimi K2.5 is an open-weight long-context model focused on agentic reasoning and tool use.

MiniMax's M2.5, a fast and inexpensive proprietary model for agentic coding, tool use, and high-volume production.

MiniMax's M2.7 agentic productivity model for coding, office workflows, tool use, and low-cost long-context execution.

MiniMax's still-active M-series model for coding, tool use, and office-style agent workflows.

Mistral's new open-weight small model for efficient long-context assistants and coding support.

Still-available OpenAI reasoning model for difficult analysis, retained as a legacy reference after GPT-5 became the default recommendation.

OpenAI's highest-capability deep research model for long, source-heavy investigations over web and private data.

Cost-efficient OpenAI reasoning model retained as a legacy API reference after GPT-5 mini became the newer default direction.

Lower-cost OpenAI deep research model for source-heavy investigations when throughput and budget matter.

OpenAI's Apache 2.0 open-weight model for local PII detection and redaction workflows.

Alibaba's older Qwen 3.6 Max Preview for agentic coding, tool use, and 262K-context reasoning.

Alibaba's older 262K-context Qwen API tier, scheduled to retire in favor of Qwen3.7 Max.

Alibaba's Qwen3.5 generation extends the Qwen line with stronger open-weight reasoning and coding performance.

Alibaba's proprietary 1M-context Qwen flagship for reasoning, coding, productivity, and long-horizon agents.

Anthropic's most capable widely released Claude model for long-running agents and high-stakes software work.

Fast and efficient Claude tier for latency-sensitive assistant and automation workloads.

Anthropic's limited-availability Mythos model for approved Project Glasswing defensive cybersecurity workflows.

Anthropic's gated research-preview frontier model for defensive cybersecurity, autonomous coding, and long-running agents.

Superseded Claude 4.6 snapshot for high-difficulty reasoning, coding, and long-running agent workflows.

Anthropic's April 2026 premium Claude model, now superseded by Opus 4.8 but still useful for pinned deployments.

Anthropic's May 2026 premium Claude model for long-horizon agentic coding, complex reasoning, and high-autonomy work.

Previous balanced Claude Sonnet snapshot, now mainly useful for pinned deployments and historical comparison.

Anthropic's previous balanced Sonnet model, retained for pinned deployments and behavior comparison.

Anthropic's balanced Sonnet model for agentic coding, tool use, and production assistants.

OpenAI's Responses API computer-use model for supervised browser and interface-control workflows.

Stable Gemini 2.5 Flash route balancing multimodal capability, latency, and production cost.

Google's stable 2.5-era native-audio Live API model for realtime multimodal voice agents.

Stable budget Gemini 2.5 tier for large-scale assistant and automation workloads.

Google's older Gemini 3 preview Flash route, now superseded by stable Gemini 3.5 Flash for most new fast-model work.

Google's low-latency Gemini 3.1 live model for realtime audio-to-audio and multimodal dialogue.

Google's stable low-cost Gemini 3.1 Flash-Lite tier for high-throughput multimodal assistant and automation workloads.

Google's current premium Gemini 3.1 preview model for multimodal reasoning, coding, long-context analysis, and agentic workflows.

Google's stable Gemini 3.5 Flash model for fast frontier multimodal, coding, and long-horizon agent workflows.

Google DeepMind's robotics-tuned Gemini for embodied reasoning, spatial planning, and physical agent tasks.

Long-context multimodal model retained as a legacy reference after retirement from ChatGPT defaults.

Widely deployed multimodal model kept as a legacy reference after retirement from ChatGPT defaults.

Lower-cost GPT-4o API tier for high-volume text-plus-image assistant and automation workloads.

xAI's 1M-context multimodal preview lane for research and multi-agent experiments alongside the Grok 4.5 flagship.

xAI's lower-cost 1M-context route for general reasoning and agentic work, now sitting behind the Grok 4.5 flagship.

xAI's current flagship for coding, agentic knowledge work, and tool-using reasoning, with a 500K context window.

xAI's lower-cost public-beta coding API model for fast agentic development, now separate from Grok Build's Grok 4.5 default.

Moonshot AI's Kimi K2.6, a 1T-parameter MoE open-weight model for long-horizon coding and agentic workflows.

Moonshot AI's thinking-only multimodal open-weight model for long-horizon coding agents and tool use.

Meta's larger open-weight Llama 4 MoE model for multimodal assistants and controlled deployments.

Meta's efficiency-focused Llama 4 MoE model with a headline 10M-token context window.

MiniMax's 1M-context native multimodal open-weight model for coding and long-horizon agents.

Mistral's flagship open-weight European multimodal model with long context and competitive enterprise API economics.

Mistral's dense 128B Medium 3.5, a frontier-class multimodal model unifying chat, reasoning, and coding behavior.

Mistral's document model for structured extraction, bounding boxes, and confidence-aware pipelines.

Meta's multimodal reasoning model powering Meta AI, with tool use, parallel-agent thinking, voice, and Muse Image integration.

Alibaba's open-weight 27B Qwen 3.6 model for agentic coding, vision-language work, and self-hosted deployment.

Previous premium GPT Image tier retained for compatibility after GPT Image 2 became the current high-quality route.

OpenAI's current state-of-the-art GPT Image model behind ChatGPT Images 2.0 and API image generation.

Lower-cost GPT Image tier for product teams that need image generation at higher volume.

xAI's standard Grok Imagine image model for API image generation and editing workflows.

Older Grok Imagine image-generation model ID retained as a legacy reference for pre-refresh integrations.

xAI's Grok Imagine Quality Mode image model for higher-realism, stronger text rendering, and brand-controlled visuals.

Meta's agentic image model for generation, editing, multi-reference composition, and tool-assisted refinement.

Google's stable Gemini 3.1 Flash Image workhorse for efficient image generation and editing.

Google's fastest and cheapest Gemini 3.1 Flash image route for high-volume 1K generation.

Google's stable Gemini 3 Pro Image model for professional visual assets and grounded image generation.

Google's preview Gemini Omni API model for multimodal video generation and conversational video editing.

xAI's Grok Imagine video model for text-to-video, image-to-video, video editing, and short creative clips.

Older Grok Imagine video-generation model ID retained as a legacy reference for pre-refresh integrations.

Runway's current flagship video model for text-to-video and image-to-video creative generation.

Google's active Veo 3.1 preview family for video generation with native audio and stronger control.

ElevenLabs' generally available expressive text-to-speech model for premium voice and dialogue output.

Google's 2.5 Pro TTS preview tier for natural, steerable one-way speech generation.

Google's expressive Gemini 3.1 Flash TTS preview, an API-first text-to-speech model spanning 70+ languages.

Google's realtime speech-to-speech translation model for the Gemini Live API.

Lower-cost OpenAI speech-to-text tier for high-volume transcription pipelines.

OpenAI text-to-speech model for responsive, API-first voice output workflows.

OpenAI speech-to-text model tier for production transcription and voice pipeline workflows.

OpenAI's earlier realtime voice model for audio-in, audio-out agents, now superseded by GPT-Realtime-2 for new flagship voice work.

OpenAI's GPT-5-class realtime voice model for reasoning, tool-using speech agents, and live support workflows.

OpenAI's realtime speech-to-speech translation model for live multilingual audio experiences.

OpenAI's streaming speech-to-text model for low-latency realtime transcription.

Earlier Google music generation route retained as reference while newer Lyria 3 surfaces take the spotlight.

ElevenLabs' current speech-to-text model line for accurate batch transcription and low-latency realtime STT.

Google's first multimodal Gemini embedding model, mapping text, images, audio, video, and PDFs into a unified vector space.