Main & Side Model List

Everything on this page comes live from the API (GET /v1/models and GET /v1/pools) โ€” the same data you get when you call it yourself. It re-syncs every hour.

Why does /v1/models differ from what the site shows?

  • Main models are the ones you can name directly in the model field. The entries with tier "main" in GET /v1/models are exactly these; each call uses main-quota points at the listed multiplier.
  • Side models are picked by Auto from the capability pools below, based on the task. Text side models are not listed in /v1/models and do not use main-quota points. Vision, image and video models can be named directly, so those do appear in /v1/models.
  • Every response carries X-TT-Model and X-TT-Tier (main / side) headers telling you which model actually served it and which tier it belongs to.
  • On the free plan, naming a main model with a multiplier of 10ร— or more is answered by Auto instead; those are marked "Paid plans" below.
# Models you can name directly (main + selectable side)
curl https://tokentable.asia/v1/models

# Side-model pools Auto uses
curl https://tokentable.asia/v1/pools

# See which model served this request
curl -si https://tokentable.asia/v1/chat/completions \
  -H "Authorization: Bearer tt-live-..." \
  -H "Content-Type: application/json" \
  -d '{"model":"auto","messages":[{"role":"user","content":"hi"}]}' \
  | grep -i '^x-tt-'

Main models (45, selectable)

Put a model ID from the table into the model field to use it. The multiplier is how many main-quota points one call uses; leave it out or send "auto" to let the router choose.

Anthropic

Model IDDescriptionContextMultiplier
claude-fable-5Most capable (selectable)Paid plans1M10ร—
claude-fable-5-1Most capable (latest, selectable)Paid plans1M10ร—
claude-opus-4-8Flagship reasoning1M6ร—
claude-opus-5Latest flagship reasoning (selectable)1M6ร—
claude-sonnet-4-6Fast + intelligent1M4ร—
claude-sonnet-5Sonnet flagship (selectable)1M4ร—

OpenAI

Model IDDescriptionContextMultiplier
gpt-5.5General purpose1M8ร—
gpt-5.6GPT flagship โ€” Sol (selectable)1M5ร—
gpt-5.6-lunaGPT flagship โ€” Luna (cost-efficient, selectable)1M1ร—
gpt-5.6-terraGPT flagship โ€” Terra (balanced, selectable)1M4ร—
gpt-6-astraGPT-6 flagship (selectable)Paid plans1M10ร—

xAI

Model IDDescriptionContextMultiplier
grok-4.20-0309-non-reasoningGrok next-gen (selectable)1M2ร—
grok-4.20-0309-reasoningGrok next-gen reasoning (selectable)1M3ร—
grok-4.3Real-time web256k2ร—
grok-4.5Grok reasoning (selectable)256k4ร—
grok-4.6Grok flagship (new)500k4ร—

Google

Model IDDescriptionContextMultiplier
gemini-3.5-flashGemini flagship1M3ร—
gemini-3.6-flashGemini next-gen (selectable)1M3ร—
gemini-3.8-flashGemini latest (selectable)1M3ร—

Mistral AI

Model IDDescriptionContextMultiplier
mistral-large-latestMistral Large (EU)256k2ร—

Together AI

Model IDDescriptionContextMultiplier
deepseek-ai/DeepSeek-V4.1-FlashDeepSeek V4.1 Flash (selectable)1M2ร—
zai-org/GLM-5.3-FlashZhipu GLM Flash (selectable)1M1ร—

Alibaba Cloud

Model IDDescriptionContextMultiplier
qwen-flashQwen fast (low-latency)1M1ร—
qwen-plusQwen classic flagship1M1ร—
qwen3-maxCN max (selectable)256k2ร—
qwen3.5-plusQwen3.5 Plus vision-language (selectable)1M2ร—
qwen3.6-plusCN flagship1M1ร—
qwen3.7-maxCN agent flagship (new)1M3ร—
qwen3.7-plusCN vision flagship (new)1M2ร—
qwen3.8-maxCN flagship (new, vision)1M3ร—
deepseek-v4-flashDeep reasoning Flash (new)1M1ร—
deepseek-v4-proDeep reasoning Pro128k1ร—
kimi-k2-thinkingKimi K2 thinking (selectable)256k2ร—
kimi-k2.5Kimi K2.5 coding (selectable)256k2ร—
kimi-k2.6Coding flagship128k1ร—
kimi-k2.7-codeKimi K2.7 coding (new)256k2ร—
kimi/kimi-k3Kimi K3 flagship (selectable)1M5ร—
glm-5.1Zhipu GLM (selectable)1M2ร—
glm-5.2Zhipu GLM flagship (new)1M2ร—
ZHIPU/GLM-5.3Zhipu GLM next-gen (selectable)1M2ร—
MiniMax-M2.5CN text (selectable)1M1ร—
MiniMax/MiniMax-M2.7CN agentic (MiniMax M2.7)1M2ร—
MiniMax/MiniMax-M3CN flagship (MiniMax M3, MSA)1M2ร—
unisound/unisound-u2Unisound U2 agent flagship (selectable)160k2ร—
xiaomi/mimo-v2.5-proXiaomi MiMo flagship (new)1M1ร—

Side models (picked by Auto)

Side models are grouped by what you want to do. With Auto (or no model set), the router picks from the matching pool. Current representative models for each pool are listed below.

Chat / Writing

Everyday chat, writing, summaries

qwen3.7-flashqwen-turboqwen-flashqwen-plus

Coding / Debug

Code generation, bug fixes, code review

deepseek-v3.2deepseek-v3qwen3-coder-flashqwen3-coder-plusqwen-coder-plus

Reasoning

Complex logic, multi-step reasoning, deep analysis

qwen3.5-flashdeepseek-r1

Long context

Very long documents, whole books, large data

qwen-long

Translation

Translation between languages

qwen3.7-flashqwen-turbo

Productivity

Document processing, reports, office work

qwen-plusglm-5.1

Vision / OCR

Image understanding, screenshot Q&A, OCR (selectable)

qwen3-vl-plusqwen3-vl-flashqwen-vl-ocr

Image generation

Text-to-image illustrations and product shots (selectable)

wan2.7-imagekling/kling-v3-omni-image-generationqwen-image-3.0

Video generation

Text or image to short video (selectable)

wan2.6-t2vkling/kling-v3-omni-video-generationhappyhorse-1.1-t2v

Voice generation

Text-to-speech

cosyvoice-v2

Pool contents change with provider availability. Image, video and voice generation use a fixed number of points per call; some advanced image/video models need to be enabled first โ€” sign in to see them in your dashboard.

This list re-syncs with the API every hour. For your own usage and enabled models, go to the dashboard