Catalogue
Models
One key, one balance. Open a model to see its pricing, its identifier and where it stands.
- Fable 5.1The top of the table: the most capable model we serve.$6.00/M in$30.00/M out
- Fable 5The previous top model, at the Fable 5.1 input and output rates.$6.00/M in$30.00/M out
- Opus 5.5The current Opus: long reasoning and agent work, at a lower rate than Opus 5.$2.40/M in$12.00/M out
- Opus 5The previous Opus: long reasoning and large contexts.$3.00/M in$15.00/M out
- Opus 4.8The last Opus 4 release, at the Opus 5 rate.$3.00/M in$15.00/M out
- Opus 4.7Kept for projects pinned to it.$3.00/M in$15.00/M out
- Opus 4.6Kept for projects pinned to it.$3.00/M in$15.00/M out
- Sonnet 5.5The current Sonnet: the workhorse of an agent session.$1.20/M in$6.00/M out
- Sonnet 5The previous Sonnet, at the Sonnet 5.5 rates.$1.20/M in$6.00/M out
- Sonnet 4.6An earlier Sonnet generation, at its own rate.$1.80/M in$9.00/M out
- Haiku 5.5The current Haiku: cheap and fast for bulk work. A prompt over 100K tokens bills at five times the rate.$0.06/M in$0.30/M out
- Haiku 4.5The previous Haiku, kept for projects pinned to it.$0.60/M in$3.00/M out
- GPT-6 AstraThe flagship of the GPT-6 generation and the most capable GPT model we serve.$6.00/M in$30.00/M outOpenAI
- GPT-6.1 SolThe newest Sol: an upgrade to GPT-6 Sol at the same input and output rates.$1.20/M in$6.00/M outOpenAI
- GPT-6.1 TerraThe mid-size GPT-6.1 model, at half the GPT-6.1 Sol rate.$0.60/M in$3.00/M outOpenAI
- GPT-6 SolThe previous Sol, at the GPT-6.1 Sol input and output rates.$1.20/M in$6.00/M outOpenAI
- GPT-6 LunaThe light GPT-6 model, at the lowest text rate in the catalogue, for bulk work.$0.06/M in$0.30/M outOpenAI
- GPT-5.6 SolA full-size GPT-5.6 model for general work.$1.20/M in$6.00/M outOpenAI
- GPT-5.6 TerraThe other full-size GPT-5.6 model, at its own output rate.$1.20/M in$7.20/M outOpenAI
- GPT-5.6 LunaThe light GPT-5.6 model, for bulk work.$0.12/M in$0.72/M outOpenAI
- GPT Image 2.5 FlareThe fast tier of GPT Image 2.5: roughly a quarter of the tokens per frame of GPT-5.4 Image 2.$7.20/M in$27.00/M outOpenAI
- GPT Image 2.5 SunburstThe precise tier of GPT Image 2.5: slower than Flare, steadier across an edit chain, same rate.$7.20/M in$27.00/M outOpenAI
- GPT-5.4 Image 2Image generation and editing, metered in output tokens.$7.20/M in$27.00/M outOpenAI
- GPT-5 ImageThe larger GPT image model, at its own output rate.$9.00/M in$36.00/M outOpenAI
- GPT-5 Image MiniThe smallest GPT image model and the cheapest GPT frame, for drafts and bulk work.$2.25/M in$7.20/M outOpenAI
- Nano Banana ProGemini 3 Pro Image: the most detailed frames, at the highest output rate.$1.80/M in$108.00/M outGoogle
- Nano Banana 2Gemini 3.1 Flash Image: fast frames at half the Pro rate.$0.45/M in$54.00/M outGoogle
- Nano Banana 2.1Frames at half the Nano Banana 2 output rate, with a higher input rate.$1.35/M in$27.00/M outGoogle
- Nano Banana 2 LiteThe light Flash build: frames at half the Nano Banana 2 rate, with less detail.$0.225/M in$27.00/M outGoogle
- Nano BananaGemini 2.5 Flash Image, the previous generation, kept for projects pinned to it.$0.27/M in$27.00/M outGoogle
- Gemini 3.8 FlashThe most capable Flash model from Google.—Google
- Gemini 3.5 Flash LiteA fast, low-cost model from Google.—Google
- DeepSeek V4 ProOpen-weight reasoning model.—DeepSeek
- Llama 4 MaverickOpen-weight model from Meta.—Meta
- Mistral Large 4European model for reasoning and code.—Mistral AI
- Grok 4.7The flagship model from xAI.—xAI
- GPT-5.5An earlier GPT-5 release from OpenAI.—OpenAI
Input and output in USD per 1M tokens; cache write and cache read are on each model’s page. For image models the output rate is per million image tokens, and the count a frame takes depends on its shape. A dash means the model is being connected and has no price yet.
Tell us which one →One key for all of them
One key, one balance, one gateway. As models are connected, the same key and the same balance reach them; the route each one answers on is named on its page.
Get API key