Skip to navigation

Models & Pricing

Every model, its limits, and its exact price

Chat models

ModelBest forContext windowMax outputInput $/1MCached input $/1MOutput $/1M
deepseek-v4-proHardest reasoning tasks1,000,000384,000$0.392$0.128$0.783
deepseek-v4.1-flashNewest DeepSeek Flash, fast and cheap (promotional pricing)1,048,576384,000$0.13$0.003$0.50
deepseek-v4-flashFast, cheap default workhorse1,000,000384,000$0.081$0.007$0.162
deepseek-v4-flash-0731Pinned V4-Flash snapshot1,048,576384,000$0.20$0.007$0.50
deepseek-v3.2Solid general model128,000128,000$0.186—$0.280
kimi-k2.7-codeCoding and agentic coding262,144262,144$0.639$0.152$3.150
kimi-k2.6General and long-form262,144262,144$0.581$0.128$2.448
kimi-k3Long-context reasoning and agents1,048,576262,144$2.50$0.30$14.00
glm-5.3General reasoning and agentic tasks1,048,573131,072$1.17$0.245$3.96
gpt-5.6-solFrontier flagship1,000,000128,000$2.40$0.20$14.00
gpt-5.6-terraFrontier, balanced cost1,000,000128,000$1.80$0.20$10.00
gpt-5.6-lunaFrontier, budget1,000,000128,000$0.18$0.02$1.10
gpt-6-astraGPT-6 flagship, hardest work (promotional pricing)1,000,000128,000$8.00$1.00$30.00
gpt-6-solGPT-6, complex coding and professional tasks (promotional pricing)1,000,000128,000$1.80$0.20$8.00
gpt-6-lunaGPT-6, fast high-volume everyday work1,000,000128,000$0.10$0.01$0.40
gpt-6.1-solGPT-6.1, newest flagship for complex coding and professional tasks1,000,000128,000$2.00$0.10$10.00
claude-opus-5-5Claude Opus 5.5, deepest reasoning and agentic work1,000,000128,000$4.00$0.12$20.00
claude-sonnet-5-5Claude Sonnet 5.5, balanced quality and cost1,000,000128,000$2.00$0.12$10.00
claude-haiku-5-5Claude Haiku 5.5, fast and cheap1,000,000128,000$0.10$0.006$0.50
claude-fable-5.1Claude Fable 5.1, frontier-tier Mythos-class model1,000,000128,000$10.00$0.15$50.00

Prices are per million tokens, input and output priced separately. Where a model has a cached-input price, the input tokens the provider reports as cache hits bill at that rate instead of the input rate; the receipt shows the split. deepseek-v3.2 has no cached tier.

The Claude models cache prompts automatically (no cache_control needed) and report cache writes as well as reads. Prompt tokens written to the cache on a turn bill at a cache-write price instead of the input price — claude-opus-5-5 3.00,‘claude−sonnet−5−5‘3.00, `claude-sonnet-5-5` 1.50, claude-haiku-5-5 0.075,‘claude−fable−5.1‘0.075, `claude-fable-5.1` 7.50 per million — and GET /v1/models reports it as cache_write_per_million_usd. Every other model has no cache-write price (null): written prompt tokens simply bill at the input price.

Prices marked promotional are introductory and can change; GET /v1/models reports promotional_pricing: true on them. Occasionally a model runs a short free window — GET /v1/models then shows a free_until timestamp on it, and requests completed before that moment are not charged.

The GPT models (gpt-5.6-sol, gpt-5.6-terra, gpt-5.6-luna, gpt-6-astra, gpt-6-sol, gpt-6-luna, and gpt-6.1-sol) and the Claude models (claude-opus-5-5, claude-sonnet-5-5, claude-haiku-5-5, and claude-fable-5.1) serve /v1/responses in addition to chat completions. The other chat models serve chat completions only.

gpt-6.1-sol always reasons, and its provider serves function tools only on the Responses surface — so a chat-completions request with tools is served through /v1/responses for you and comes back as an ordinary chat completion (the same way max is handled; see the table below). temperature other than 1 is rejected on it. Image inputs on the Claude models must be inline data: URLs; remote image URLs are rejected.

Requests that omit the output-limit field get a per-model default: 4,096 tokens for deepseek-v3.2, 8,192 tokens for every other chat model listed above.

Reasoning effort by model

Every chat model takes the OpenAI reasoning_effort field (reasoning.effort on /v1/responses) to set its thinking level. The accepted values differ per model. A value the model does not accept is rejected with a 400 — Invalid request. when the platform catches it, The selected model could not complete the request. when the provider does — and nothing is charged either way. Verified against production on 2026-09-24; the gpt-6.1-sol and Claude rows on 2026-10-21.

ModelAccepted valuesField omittedNotes
deepseek-v4-pro, deepseek-v4.1-flash, deepseek-v4-flash, deepseek-v4-flash-0731none, minimal, low, medium, highreasoning offminimal and above turn reasoning on; none keeps it off
deepseek-v3.2none, minimal, low, medium, highreasoning offonly medium and high produce reasoning; none, minimal, and low behave like omitting the field
kimi-k2.7-code, kimi-k2.6none, low, medium, highreasoning onnone turns reasoning off; minimal is rejected
kimi-k3none, minimal, low, medium, high, xhigh, maxreasoning onnone turns reasoning off
glm-5.3low, medium, highreasoning onlow turns reasoning off (zero reasoning tokens); none and minimal are rejected
gpt-5.6-sol, gpt-5.6-terra, gpt-5.6-lunanone, low, medium, high, xhigh on chat completions; those plus max on /v1/responsesprovider defaultnone turns reasoning off; minimal is rejected on both endpoints
gpt-6-sol, gpt-6-lunanone, low, medium, high, xhigh, max on both endpointsprovider defaultnone turns reasoning off; minimal is rejected; max on chat completions is served through the Responses surface for you
gpt-6-astralow, medium, high, xhigh, max on both endpointsprovider defaultalways reasons — none and minimal are rejected; max on chat completions is served through the Responses surface for you
gpt-6.1-sollow, medium, high, xhigh, max on both endpointsprovider defaultalways reasons — none and minimal are rejected; max on chat completions is served through the Responses surface for you; temperature other than 1 is rejected
claude-opus-5-5, claude-sonnet-5-5, claude-haiku-5-5any value is accepted (none, minimal, low, medium, high, xhigh, max)reasoning on (high)the provider does not validate the value; max produces the most reasoning tokens
claude-fable-5.1any value is accepted (none, minimal, low, medium, high, xhigh, max)reasoning offthe provider does not validate the value; set an effort to turn reasoning on

Reasoning tokens are billed as output tokens at the model’s output rate and are included in usage.completion_tokens; the receipt shows the exact charge. The DeepSeek, Kimi, and GLM models return their thinking as message.reasoning_content (and delta.reasoning_content in streams) next to the normal content. The GPT models never return reasoning text — only the count in usage.completion_tokens_details.reasoning_tokens. The Claude models return reasoning text inconsistently (sometimes message.reasoning / reasoning_content, sometimes nothing) while always billing the reasoning tokens; do not depend on the text being present.

Image generation models

Four models serve POST /v1/images/generations. Image requests are never billed by tokens — the price is fixed by the request itself and the receipt settles at exactly that amount.

FLUX — billed per megapixel

ModelBest forPrice
flux-1-schnellFast, cheap generations$0.002 per megapixel
flux-pro-1.1Higher-fidelity generations$0.04 per megapixel

FLUX models accept any WIDTHxHEIGHT size with each side 256–2560 px and an area up to 4 megapixels, billed as ceil(width × height / 2²⁰) megapixels — a 1024x1024 image is 1 MP; 1600x1024 rounds up to 2 MP.

GPT Image — billed per image (quality × size)

gpt-image-2

Quality1024×10241024×15361536×1024
low$0.005$0.004$0.004
medium$0.043$0.031$0.031
high$0.151$0.125$0.125

gpt-image-1.5

Quality1024×10241024×15361536×1024
low$0.008$0.009$0.009
medium$0.024$0.040$0.040
high$0.113$0.100$0.100

Omitted size and quality default to 1024x1024 and high. Requesting n images multiplies the per-image price by n (up to 4). See Image Generation for the full parameter reference.

Both GPT Image models also serve POST /v1/images/edits, sharing this same per-image (quality × size) grid: an edit costs exactly what a generation of the same quality, size, and n costs. FLUX models do not support editing.

GET /v1/models is the authoritative live list, including current pricing. Prices on this page are current as of 2026-09-24. See Models & Receipts for the endpoint reference.