Skip to content
IQ

Models

IQ supports the chat models below, grouped by provider. Pass any listed id directly as model in a request, or use a routing alias instead -- see Routing for how aliases pick a model.

This page is a snapshot. Provider rosters change as providers ship new versions, so the authoritative, always-current list comes from the live endpoint:

curl https://gateway.iq-routing.com/v1/models \
  -H "Authorization: Bearer gw_live_xxxxxxxx"

Treat the sections below as a reference and GET /v1/models as the source of truth.

Model id prefixes

A model id from one of the providers below takes the form <prefix>/<the provider's own model id>, for example deepseek/deepseek-flash or gemini/gemini-3.8-flash. A custom endpoint (see Providers) uses custom/<your-endpoint-slug> instead. Anthropic and OpenAI ids also work in the unprefixed form shown in their sections below (for example claude-opus-5-5, gpt-5-mini); the anthropic/ and openai/ prefixed form works too.

| Provider | Prefix | |---|---| | OpenAI | openai/ | | Anthropic | anthropic/ | | Google | gemini/ | | Mistral | mistral/ | | xAI | xai/ | | Cohere | cohere/ | | DeepSeek | deepseek/ | | Z.ai (GLM) | zai/ | | Moonshot AI (Kimi) | moonshot/ | | Qwen (Alibaba Cloud) | dashscope/ | | MiniMax | minimax/ | | Together AI | together_ai/ | | Fireworks AI | fireworks_ai/ | | DeepInfra | deepinfra/ | | Replicate | replicate/ | | OpenRouter | openrouter/ | | Groq | groq/ | | Cerebras | cerebras/ |

AWS Bedrock

AWS Bedrock has no separate model ids in IQ -- request Claude models by their usual ids. With a Bedrock key connected and Bedrock routing turned on (off by default) from /providers, IQ can serve claude-haiku-4-5-20251001, claude-sonnet-5, and claude-opus-5 as compatibility models through your Bedrock account, where your AWS account has them enabled, in a supported region.

Anthropic

Current Claude lineup:

  • Haiku 5.5 (claude-haiku-5-5)
  • Sonnet 5.5 (claude-sonnet-5-5)
  • Opus 5.5 (claude-opus-5-5)
  • Fable 5.1 (claude-fable-5-1)

Availability depends on your connected provider and organization settings. Older accepted IDs are compatibility aliases. They do not describe the current lineup:

  • claude-haiku-4-5-20251001
  • claude-opus-4-6
  • claude-opus-4-7
  • claude-opus-4-8
  • claude-opus-5
  • claude-sonnet-4-6
  • claude-sonnet-5

Name a current Claude ID to request that model. Haiku 5.5 has no model fallback: if it cannot serve the request, the request fails. Sonnet 5.5, Opus 5.5 and Fable 5.1 can fall back within Anthropic when needed. Your organization's settings and subagent routing choices still apply. See Routing.

The older claude-fable-5 ID is also accepted for Anthropic-family routing. See capability aliases for cap:orchestrate, the way to request that older Fable model itself, with its own limits and Anthropic's data-retention requirement.

Cerebras

Cerebras provides hosted access to open-weight chat models.

  • cerebras/gpt-oss-120b
  • cerebras/qwen-3.8-27b

Cohere

Cohere provides its own chat model family through the gateway.

  • cohere/command-r-plus-08-2024
  • cohere/command-r-08-2024
  • cohere/command-r7b-12-2024

DeepInfra

DeepInfra provides hosted access to open-weight chat models.

  • deepinfra/deepseek-ai/DeepSeek-V4.1-Flash
  • deepinfra/deepseek-ai/DeepSeek-V4-Pro-0813
  • deepinfra/deepseek-ai/DeepSeek-V4-Flash-0731
  • deepinfra/zai-org/GLM-5.3
  • deepinfra/moonshotai/Kimi-K3
  • deepinfra/MiniMaxAI/MiniMax-M3
  • deepinfra/openai/gpt-oss-120b
  • deepinfra/meta-llama/Llama-3.3-70B-Instruct-Turbo

DeepSeek

DeepSeek provides its own chat model family through the gateway.

  • deepseek/deepseek-flash
  • deepseek/deepseek-v4-pro

Fireworks AI

Fireworks AI provides hosted access to open-weight and partner chat models.

  • fireworks_ai/accounts/fireworks/models/deepseek-v4p1-flash
  • fireworks_ai/accounts/fireworks/models/glm-5p3
  • fireworks_ai/accounts/fireworks/models/glm-5p3-flash
  • fireworks_ai/accounts/fireworks/models/kimi-k3
  • fireworks_ai/accounts/fireworks/models/minimax-m3
  • fireworks_ai/accounts/fireworks/models/qwen3p8-max
  • fireworks_ai/accounts/fireworks/models/gpt-oss-120b

Google

Google provides its Gemini chat model family through the gateway.

  • gemini/gemini-3-flash-preview
  • gemini/gemini-3.1-flash-lite
  • gemini/gemini-3.1-pro-preview
  • gemini/gemini-3.5-flash-lite
  • gemini/gemini-3.8-flash

Groq

Groq provides hosted access to open-weight chat models.

  • groq/openai/gpt-oss-120b
  • groq/openai/gpt-oss-20b
  • groq/qwen/qwen3.8-27b

MiniMax

MiniMax provides its own chat model family through the gateway.

  • minimax/MiniMax-M3

Mistral

Mistral provides its own chat model family through the gateway.

  • mistral/mistral-medium-3-5
  • mistral/mistral-large-2512
  • mistral/mistral-small-2603
  • mistral/ministral-14b-2512
  • mistral/codestral-2508
  • mistral/zai-glm-5-3

Moonshot AI (Kimi)

Moonshot AI provides its Kimi chat model family through the gateway.

  • moonshot/kimi-k3
  • moonshot/kimi-k2.7-code

OpenAI

OpenAI's chat models are available through the gateway:

  • gpt-4o
  • gpt-4o-mini
  • gpt-5
  • gpt-5-mini
  • gpt-5-nano
  • gpt-5.5
  • gpt-5.6-luna
  • gpt-5.6-sol
  • gpt-5.6-terra
  • gpt-6-astra
  • gpt-6-luna
  • gpt-6.1-sol
  • gpt-6-sol
  • o3
  • o3-mini

gpt-6-astra, gpt-6.1-sol, and gpt-6-sol are exact pins: naming any directly serves exactly that model whenever it can serve the request, falling back to another OpenAI model only if it can't. The current Claude ids, claude-haiku-5-5, claude-sonnet-5-5, claude-opus-5-5 and claude-fable-5-1, are exact pins too; their fallback behavior is described under Anthropic. See Routing for how exact pins differ from the family-lock behavior of every other id on this page.

GPT-6.1 Sol

Pass model="gpt-6.1-sol" or model="openai/gpt-6.1-sol" to request GPT-6.1 Sol. Naming gpt-6-sol continues to request GPT-6 Sol. Your organization's model permissions still apply. The x-iq-served-model response header appears only when routing substitutes or changes the model; to see which model answered, find the request in Requests by its X-Request-Id.

Use IQ's /v1/chat/completions endpoint for text, tool calls, image inputs, and structured outputs with GPT-6.1 Sol. Its context window is 1,050,000 tokens, with up to 128,000 output tokens. Supported reasoning_effort values are low, medium, high, xhigh, and max. If you name GPT-6.1 Sol and pass none or minimal, IQ uses low.

Standard token rates in USD per million tokens:

  • Up to 272,000 input tokens: $2.00 input, $0.10 cached input, $2.50 cache writes, and $10.00 output.
  • More than 272,000 input tokens: $4.00 input, $0.20 cached input, $5.00 cache writes, and $15.00 output.

Above 272,000 input tokens, the higher rates apply to the whole request. Reasoning tokens are charged at the output rate. Actual cost depends on token usage and provider-reported cache usage; these token rates do not describe your IQ subscription.

Sources: OpenAI model page and OpenAI API pricing, checked October 3, 2026.

OpenRouter

OpenRouter aggregates chat models from multiple providers behind a single API.

  • openrouter/deepseek/deepseek-v4.1-flash
  • openrouter/deepseek/deepseek-v4-flash-0731
  • openrouter/z-ai/glm-5.3
  • openrouter/z-ai/glm-5.3-flash
  • openrouter/moonshotai/kimi-k3
  • openrouter/minimax/minimax-m3

Qwen (Alibaba Cloud)

Alibaba Cloud provides its Qwen chat model family through the gateway.

  • dashscope/qwen3.8-max
  • dashscope/qwen3.8-flash
  • dashscope/qwen3.7-plus

Replicate

Replicate provides hosted access to open-weight and partner chat models. IQ never picks these models on its own. They're used only where you choose them: when you name one directly, add it to your own custom table, or put it in a team's allow-list. They don't support tool calling.

  • replicate/meta/llama-4-scout-instruct
  • replicate/meta/llama-4-maverick-instruct
  • replicate/openai/gpt-oss-120b
  • replicate/deepseek-ai/deepseek-v3.1
  • replicate/qwen/qwen3-235b-a22b-instruct-2507
  • replicate/moonshotai/kimi-k2.6

Together AI

Together AI provides hosted access to open-weight chat models.

  • together_ai/deepseek-ai/DeepSeek-V4.1-Flash
  • together_ai/deepseek-ai/DeepSeek-V4-Pro-0813
  • together_ai/deepseek-ai/DeepSeek-V4-Flash-0731
  • together_ai/zai-org/GLM-5.3
  • together_ai/moonshotai/Kimi-K3
  • together_ai/MiniMaxAI/MiniMax-M3
  • together_ai/openai/gpt-oss-120b
  • together_ai/meta-llama/Llama-3.3-70B-Instruct-Turbo

xAI

xAI provides its Grok chat model family through the gateway.

  • xai/grok-4.7
  • xai/grok-4.3
  • xai/grok-build-0.1

Z.ai (GLM)

Z.ai provides its GLM chat model family through the gateway.

  • zai/glm-5.3
  • zai/glm-5.3-flash

Next

  • Providers -- add an API key for each provider you want to use.
  • Routing -- how routing aliases pick a model.