Chat / Instruct

DeepSeek V4 Flash, an open-weight DeepSeek model available via OpenRouter.

OpenAI's open-weight 20B model in the gpt-oss family — small, cheap, and self-hostable.

Meta Llama 3.1 8B Instruct — a widely deployed small open-weight model.

Meta's Llama 3.2 1B Instruct — smallest open-weight Llama 3.2 text model.

OpenAI's GPT-6 Luna — low-cost GPT-6 tier with a 1M-token context window and vision/file input.

Alibaba's Qwen3.7 Flash — fast, low-cost model with a 1M-token context window and vision/video input.

Open-weight 117B MoE model (5.1B active) achieving near o4-mini reasoning. Apache 2.0 licensed, runs on a single 80GB GPU.

Inception's Mercury 2.5 diffusion language model — very high throughput, 260K-token context window.

Mid-size open-weight Gemma model with vision support.

Compact open-weight model for edge and mobile deployment.

Ultra-lightweight GPT-5 for high-speed, low-cost text generation.

Meta's Llama 3.2 3B Instruct — small open-weight text model for lightweight and on-device use.

Compact Qwen 3.5 for single-GPU deployment.

Upstage's Solar Mini 4 — compact, low-cost text model with a 524K-token context window.

Compact Qwen 3 for edge and single-GPU deployment. Open weight.

OpenAI's GPT-5.6 Luna — low-cost GPT-5.6 tier with vision and file input.

NVIDIA's Nemotron 3.5 Lightning — fast, low-cost text model with a 262K-token context window.

DeepSeek's V4.1 Flash — fast, low-cost model with a 1M-token context window and vision input.

Optimized for coding, reasoning, and tool use.

GLM 4.7 Flash, an open-weight Z.ai model available via OpenRouter.

IBM's Granite 4.2 8B compact open model with a 131K-token context window.

Poolside's Laguna XS 2.1 — compact coding-agent model with a 262K-token context window.

Amazon Nova Lite for fast, cost-efficient tasks.

Gemma 4 26B A4B, an open-weight Google model available via OpenRouter.

Qwen3 Coder 30B A3B Instruct, an open-weight Alibaba model available via OpenRouter.

Largest Gemma 3 model with strong reasoning and instruction following.

Mid-size Qwen 3 with strong coding and math capabilities. Open weight.

Tencent's Hunyuan Hy3 text model with a 262K-token context window.

Speed-optimized Gemini with strong reasoning and multimodal capabilities.

Google's Gemma 4 31B — largest dense open-weight Gemma 4 model, with image input and a 262K-token context window.

Poolside's Laguna S 2.1 coding-agent model with a 1M-token context window.

Unified model combining fast instruct, deep reasoning, and multimodal chat. 119B params.

Largest Qwen 3 model with hybrid thinking modes for flexible reasoning control.

Qwen3 Next 80B A3B Instruct, an open-weight Alibaba model available via OpenRouter.

Alibaba's Qwen3.8 Flash — fast, low-cost model with vision and video input.

Upstage's Solar Pro 4 text model with a 524K-token context window.

Mistral Small 3.2 24B, an open-weight Mistral AI model available via OpenRouter.

Fast and efficient Gemini model for high-throughput workloads.

Google's Gemini 2.5 Flash Lite — lowest-cost Gemini 2.5 tier with multimodal input and a 1M-token context window.

GPT-4.1 Nano, a proprietary OpenAI model available via OpenRouter.

OpenAI's GPT-6 Luna Pro — extended-reasoning variant of GPT-6 Luna with vision/file input.

Widely deployed open-weight model with strong general capabilities.

Natively multimodal MoE model with 10M context. 109B total, 17B active. Fits single H100.

Qwen3 30B A3B, an open-weight Alibaba model available via OpenRouter.

Qwen3.6 35B A3B, an open-weight Alibaba model available via OpenRouter.

Seed-2.0-Mini, an open-weight ByteDance model available via OpenRouter.

Qwen3 VL 32B Instruct, an open-weight Alibaba model available via OpenRouter.

Qwen3 VL 8B Instruct, an open-weight Alibaba model available via OpenRouter.

Z.ai's GLM 5.3 Flash — fast, low-cost model with a 1M-token context window and vision/video input.

Qwen3 14B, an open-weight Alibaba model available via OpenRouter.

Qwen3 Coder Next, an open-weight Alibaba model available via OpenRouter.

GLM 4.5 Air, an open-weight Z.ai model available via OpenRouter.

Reasoning-focused model with chain-of-thought capabilities rivaling o1.

Latest DeepSeek V3 with improved reasoning and coding. 671B MoE (37B active), MIT licensed, 164K context.

Strong reasoning and coding with 106B total, 12B active MoE architecture.

Xiaomi's MiMo V2.5 — low-cost multimodal model (text, image, audio, video input) with a 1M-token context window.

Xiaomi's MiMo V2.6 Flash — low-cost multimodal model (text, image, audio, video input) with a 1M-token context window.

Mid-size Qwen 3.5 MoE model with 35B total, 3B active parameters.

Fast and efficient Gemini 3 model for high-throughput workloads.

Cost-efficient smaller GPT-4o variant for lightweight tasks.

MiniMax general-purpose LLM with competitive reasoning and coding capabilities.

Edge-optimized model with vision support. Apache 2.0 licensed.

Qwen3 Next 80B A3B Thinking, an open-weight Alibaba model available via OpenRouter.

Qwen3 VL 30B A3B Instruct, an open-weight Alibaba model available via OpenRouter.

Qwen3.5-35B-A3B, an open-weight Alibaba model available via OpenRouter.

Alibaba's Qwen3.8 Omni Flash — low-cost omni-modal model (text, image, audio, video input) with a 1M-token context window.

DeepSeek V4 Pro, an open-weight DeepSeek model available via OpenRouter.

Qwen3 VL 8B Thinking, an open-weight Alibaba model available via OpenRouter.

Largest open Llama 4 with 128 experts. 400B total, 17B active. Beats GPT-4o on benchmarks.

Qwen3.6 Flash, an open-weight Alibaba model available via OpenRouter.

Qwen3 Coder Flash, an open-weight Alibaba model available via OpenRouter.

Qwen3.5-27B, an open-weight Alibaba model available via OpenRouter.

Ultra-lightweight GPT-5.4 for high-speed, low-cost tasks.

OpenAI's GPT-5.6 Luna Pro — extended-reasoning variant of GPT-5.6 Luna with vision/file input.

Lightweight Grok optimized for cost-efficient reasoning.

Native multimodal Qwen with text, image, and video processing.

Qwen3 VL 235B A22B Instruct, an open-weight Alibaba model available via OpenRouter.

Qwen3 VL 30B A3B Thinking, an open-weight Alibaba model available via OpenRouter.

Alibaba's Qwen3.8 27B — compact open model with vision and video input.

Latest MiniMax general-purpose LLM with improved reasoning.

Open-weight 671B MoE model with strong coding and reasoning at low cost.

Google's Gemini 3.6 Flash — fast multimodal model (text, image, audio, video, file input) with a 1M-token context window.

Google's Gemini 3.7 Flash — fast multimodal model (text, image, audio, video, file input) with a 1M-token context window.

Google's Gemini 3.8 Flash — fast multimodal model (text, image, audio, video, file input) with a 1M-token context window.

Alibaba's Qwen3 235B A22B Thinking (2507) — open-weight MoE reasoning model.

Updated DeepSeek V3 with improved coding and reasoning performance.

Gemini 3.1 Flash Lite, a proprietary Google model available via OpenRouter.

Compact GPT-5 variant for lightweight tasks and rapid prototyping.

GPT-5.1-Codex-Mini, a proprietary OpenAI model available via OpenRouter.

Seed-2.0-Lite, an open-weight ByteDance model available via OpenRouter.

DeepSeek V3.2 Exp, an open-weight DeepSeek model available via OpenRouter.

Qwen-Plus, an open-weight Alibaba model available via OpenRouter.

Qwen3.5-122B-A10B, an open-weight Alibaba model available via OpenRouter.

DeepSeek V3.1 Terminus, an open-weight DeepSeek model available via OpenRouter.

Latest Haiku tier with improved capabilities at fast speed and low cost.

MiniMax M3, an open-weight MiniMax model available via OpenRouter.

Large Qwen 3.5 MoE model with 122B total, 10B active parameters.

Google's Gemini 3.5 Flash Lite — lowest-cost Gemini 3.5 tier, multimodal input (text, image, audio, video, file) with a 1M-token context window.

GLM 4.6V, an open-weight Z.ai model available via OpenRouter.

Meituan's LongCat 2.0 text model with a 1M-token context window.

MiniMax M2, an open-weight MiniMax model available via OpenRouter.

MiniMax M2.1, an open-weight MiniMax model available via OpenRouter.

Meta's Muse Glimmer 30B — compact Muse model with image input.

Nova 2 Lite, a proprietary Amazon model available via OpenRouter.

Qwen3 Coder 480B A35B, an open-weight Alibaba model available via OpenRouter.

Qwen3.6 27B, an open-weight Alibaba model available via OpenRouter.

Qwen3.7 Plus, an open-weight Alibaba model available via OpenRouter.

Alibaba's latest flagship with 1M context and advanced agentic coding.

Mistral Small 3.1 24B, an open-weight Mistral AI model available via OpenRouter.

Z.ai's GLM 5.3 FlashX — faster GLM 5.3 Flash tier with a 1M-token context window and vision/video input.

High-capability Gemini model for complex reasoning and coding tasks.

GLM 5.2, an open-weight Z.ai model available via OpenRouter.

Open-source frontier model with 355B parameters. MIT licensed.

GPT-4.1 Mini, a proprietary OpenAI model available via OpenRouter.

Mistral Medium 3.1, an open-weight Mistral AI model available via OpenRouter.

Qwen3 VL 235B A22B Thinking, an open-weight Alibaba model available via OpenRouter.

Xiaomi's MiMo V2.6 Pro — multimodal model (text, image, audio, video input) with a 1M-token context window.

Xiaomi's MiMo V2.5 Pro text model with a 1M-token context window.

GPT-5.2 general-purpose model.

Gemini 3.5 Flash, a proprietary Google model available via OpenRouter.

Thinking Machines Lab's Inkling Small — lower-cost Inkling (text, image, audio input) with a 524K-token context window.

Open-weight multimodal model with agent swarm mode supporting up to 100 parallel sub-agents.

Largest Qwen 3.5 MoE model with 397B total, 17B active parameters.

Updated R1 with improved reasoning accuracy and reduced hallucination.

High-capability Gemini 3 model. Deprecated in favor of 3.1 Pro.

GPT-5 variant optimized for code generation and software engineering.

GPT-5.1 code-optimized variant.

Grok 4.3, a proprietary xAI model available via OpenRouter.

State-of-the-art 1T MoE model with 32B active parameters. Strong coding and agentic capabilities.

Mistral's most capable model. 675B MoE with 41B active parameters.

Qwen3.5 397B A17B, an open-weight Alibaba model available via OpenRouter.

ByteDance Seed 2.0 Code — coding-specialized Seed 2.0 model with image/video input and a 262K-token context window.

ByteDance Seed 2.1 Turbo — fast general model with image/video input and a 262K-token context window.

MiniMax M1 open-weight LLM with a 1M-token context window.

OpenAI's GPT-5.6 Terra model — mid GPT-5.6 tier with vision and file input.

GLM 4.5V, an open-weight Z.ai model available via OpenRouter.

Frontier 744B model trained on Huawei Ascend chips. Open source with strong agentic capabilities.

OpenAI's GPT-6 Sol — high-end GPT-6 tier with a 1M-token context window and vision/file input.

OpenAI's GPT-6.1 Sol — high-end GPT-6.1 tier with a 1M-token context window and vision/file input.

Kimi K2 Thinking, an open-weight Moonshot AI model available via OpenRouter.

Qwen3 Coder Plus, an open-weight Alibaba model available via OpenRouter.

Coding-focused agentic model built on Kimi K2.6 with ~30% lower thinking-token usage. 1T MoE (32B active), native INT4 quantization, 256K context, text/image/video input. Forces thinking and preserve_thinking modes; tuned for long-horizon software engineering and agent workflows.

Moonshot AI's open-weight flagship: natively multimodal 1T MoE model (32B active) with text, image, and video input, thinking and non-thinking modes, and 256K context. State-of-the-art open-source coding and agentic performance (SWE-Bench Verified 80.2); Agent Swarm scales to 300 parallel sub-agents.

GPT-5.2 code-optimized variant.

GPT-5.3 code-optimized variant.

OpenAI's latest frontier model combining reasoning, coding, and agentic workflows.

Latest GPT-5.4 code-optimized variant with industry-leading coding capabilities.

Moonshot AI's Kimi K3 model with a 1M-token context window and vision/video input.

Kwaipilot's KAT-Coder-Pro V2.5 agentic coding model with a 262K-token context window.

Compact GPT-5.4 variant balancing capability and cost.

Alibaba's highest capability Qwen 3 model.

Qwen 3 Max with extended reasoning and chain-of-thought capabilities.

Fast and affordable Claude model for high-throughput tasks.

Anthropic's Claude Sonnet 5.5 — mid-tier Claude 5.5 model with a 1M-token context window and vision/file input.

xAI's Grok 4.5 model with a 500K-token context window and vision/file input.

xAI's Grok 4.6 model with a 500K-token context window and vision input.

xAI's Grok 4.7 model with a 500K-token context window and vision/file input.

Amazon Nova Pro for balanced capability and cost.

Tencent's Hunyuan Hy4 preview text model with a 1M-token context window.

High-capability Claude model balancing intelligence and speed.

Latest Sonnet with Opus-tier capabilities at Sonnet pricing.

Anthropic's Claude Sonnet 5 — mid-tier Claude 5 model with a 1M-token context window and vision/file input.

Z.ai's GLM 5.3 open-weight model with a 1M-token context window.

Thinking Machines Lab's Inkling — multimodal model (text, image, audio input) with a 524K-token context window.

Sakana AI's Namazu model with a 262K-token context window and vision/file input.

xAI's Grok Build 0.1 coding-agent model.

Lightweight reasoning model balancing chain-of-thought rigor with speed and cost efficiency.

Coding-focused frontier model scoring 94% of Claude Opus 4.6. 744B MoE trained on Huawei Ascend 910B. #1 on SWE-Bench Pro (open source).

GLM 5 Turbo, an open-weight Z.ai model available via OpenRouter.

GLM 5V Turbo, an open-weight Z.ai model available via OpenRouter.

GPT-5, a proprietary OpenAI model available via OpenRouter.

GPT-5.1, a proprietary OpenAI model available via OpenRouter.

GPT-5.1-Codex-Max, a proprietary OpenAI model available via OpenRouter.

Latest Grok with improved instruction following and reduced hallucination.

Muse Spark 1.3 multimodal chat model — accepts text, image, audio, video and file input.

Qwen3.7 Max, an open-weight Alibaba model available via OpenRouter.

GPT-5.5, a proprietary OpenAI model available via OpenRouter.

OpenAI's GPT-5.6 Sol model — high-end GPT-5.6 tier with vision and file input.

Previous Opus generation with strong reasoning and coding.

Anthropic's most capable model with 1M token context and advanced reasoning.

Claude Opus 4.7, a proprietary Anthropic model available via OpenRouter.

Mistral Medium 3.5, an open-weight Mistral AI model available via OpenRouter.

Anthropic's Claude Opus 5.5 frontier model with a 1M-token context window and vision/file input.

Alibaba's Qwen3.8 Max flagship model with vision and video input.

Claude Opus 4.8, a proprietary Anthropic model available via OpenRouter.

Anthropic's Claude Opus 5 frontier model with a 1M-token context window and vision input.

Sakana AI's Fugu Max with a 1M-token context window and vision/file input.

Google's current flagship model with top benchmark scores and 1M context.

GPT-4.1, a proprietary OpenAI model available via OpenRouter.

OpenAI's GPT-5.6 Terra Pro — extended-reasoning variant of GPT-5.6 Terra with vision/file input.

OpenAI's GPT-6 Sol Pro — extended-reasoning variant of GPT-6 Sol with vision/file input.

OpenAI's GPT-6.1 Sol Pro — extended-reasoning variant of GPT-6.1 Sol with vision/file input.

Reasoning-focused model with step-by-step deliberation for complex math, coding, and science tasks.

Alibaba's Qwen3.8 2.4T-parameter MoE (95B active) text model with a 1M-token context window.

Cohere's latest flagship model for enterprise RAG, tool use, and agents.

Scalable enterprise model optimized for RAG and multilingual tasks.

OpenAI's flagship multimodal model with strong reasoning, coding, and vision capabilities.

Amazon's most capable LLM for complex reasoning and enterprise tasks.

Z.ai's GLM 5.3 Prime — premium GLM 5.3 tier with a 1M-token context window.

OpenAI's GPT-6 Astra frontier model with vision and file input.

Claude Sonnet 4, a proprietary Anthropic model available via OpenRouter.

Fireworks AI's Ember-1 model with a 1M-token context window and vision input.

xAI's flagship LLM trained on 200K+ GPUs with real-time web and X integration.

Claude Fable 5, a proprietary Anthropic model available via OpenRouter.

OpenAI's GPT-5.6 Sol Pro — extended-reasoning variant of GPT-5.6 Sol with vision/file input.

Alibaba's Qwen3.8 Max Prime — premium Qwen3.8 Max tier with a 1M-token context window and vision/video input.

The high-tier variant of Sakana AI's Fugu orchestration model. Reportedly outperforms Opus 4.8, GPT-5.5 and Gemini 3.1 Pro on many benchmarks (73.7 on SWE-bench Pro), standing shoulder-to-shoulder with Claude Fable 5. Built on the TRINITY and Conductor (ICLR 2026) research.

Sakana AI's Fugu Ultra v2 flagship with a 1M-token context window and vision/file input.

Anthropic's Claude Fable 5.1 frontier model with a 1M-token context window and vision input.

OpenAI's GPT-6 Astra Pro — extended-reasoning variant of the GPT-6 Astra frontier model with vision/file input.

Claude Opus 4, a proprietary Anthropic model available via OpenRouter.

Claude Opus 4.1, a proprietary Anthropic model available via OpenRouter.

GPT-5 Pro, a proprietary OpenAI model available via OpenRouter.

Most capable reasoning model in OpenAI's lineup with extended thinking for maximum reliability.

GPT-5.2 Pro, a proprietary OpenAI model available via OpenRouter.

Highest capability GPT-5.4 tier with maximum reasoning depth. Premium pricing.

GPT-5.5 Pro, a proprietary OpenAI model available via OpenRouter.

Sakana AI's multi-agent orchestration model — a trained "conductor" LM that routes and synthesizes responses across a swappable pool of underlying frontier models, exposed as a single API.

Latest Gemma generation optimized for reasoning and agentic workflows.