Model prices
One model per row, Output discounts and expandable exact prices for all token categories.
The table uses the published default Individual registration pricing template, not a signed-in visitor’s private overrides. Each model has one expandable row. The summary shows official and public Model Gate Output prices in USD per one million provider tokens, the Output discount and an integer Tokens coefficient. Expand the model for Input, Output, Cache read, Cache write and Reasoning prices. This table is not the Images tariff catalog.
Discount is (official - public) / official × 100% only when the official reference is positive and both prices are known. If the public price is higher, no discount is displayed; the two prices remain visible. Unknown references and prices are not zero and no fixed discount is assumed.
At the configured display rate 100,000,000 Tokens = R USD, the coefficient per one provider token is USD price × 100 / R. Coefficients are rounded UP to whole numbers for display only. This rounding does not change USD prices, stored tariffs or settlement: Tokens are a display equivalent, not a second charge. The former Tokens-per-million column is omitted.
Published default Individual rates, without account overrides. One row per model: prices, discount and Tokens coefficient in the summary refer to Output. Expand a model for all token categories. USD prices are per 1M provider tokens.
Display rate: 100,000,000 Tokens = 15 USD. Batch request coefficient: 0.1×. Normal synchronous/native async prices are shown; Batch is separate.
claude-fable-5Official Output$50Our Output$5Output discount90%cheaperTokens coefficient33.3333
claude-fable-5-1Official Output$50Our Output$5Output discount90%cheaperTokens coefficient33.3333
claude-haiku-5-5Official Output$2.5Our Output$0.25Output discount90%cheaperTokens coefficient1.6667
claude-opus-4-6Official Output$25Our Output$2.5Output discount90%cheaperTokens coefficient16.6667
claude-opus-4-7Official Output$25Our Output$2.5Output discount90%cheaperTokens coefficient16.6667
claude-opus-4-8Official Output$25Our Output$2.5Output discount90%cheaperTokens coefficient16.6667
claude-opus-5Official Output$25Our Output$2.5Output discount90%cheaperTokens coefficient16.6667
claude-opus-5-5Official Output$20Our Output$3Output discount85%cheaperTokens coefficient20
claude-sonnet-4-6Official Output$15Our Output$1.5Output discount90%cheaperTokens coefficient10
claude-sonnet-5Official Output$10Our Output$1Output discount90%cheaperTokens coefficient6.6667
claude-sonnet-5-5Official Output$10Our Output$1Output discount90%cheaperTokens coefficient6.6667
deepseek-v4-flashOfficial Output$0.6Our Output$0.06Output discount90%cheaperTokens coefficient0.4
deepseek-v4-proOfficial Output$3.96Our Output$0.6Output discount84.85%cheaperTokens coefficient4
deepseek-v4.1-flashOfficial Output$1.2Our Output$0.12Output discount90%cheaperTokens coefficient0.8
gemini-3.1-proOfficial Output$12Our Output$1.2Output discount90%cheaperTokens coefficient8
gemini-3.6-flashOfficial Output$4.125Our Output$1Output discount75.76%cheaperTokens coefficient6.6667
gemini-3.7-flashOfficial Output$4.125Our Output$0.5Output discount87.88%cheaperTokens coefficient3.3333
gemini-3.8-flashOfficial Output$3.75Our Output$0.5Output discount86.67%cheaperTokens coefficient3.3333
glm-5-turboOfficial Output$3.2Our Output$1Output discount68.75%cheaperTokens coefficient6.6667
glm-5.2Official Output$4.4Our Output$0.44Output discount90%cheaperTokens coefficient2.9333
glm-5.3Official Output$4.4Our Output$1Output discount77.27%cheaperTokens coefficient6.6667
glm-5.3-flashOfficial Output$0.5Our Output$0.1Output discount80%cheaperTokens coefficient0.6667
gpt-5.6-lunaOfficial Output$0.4Our Output$0.04Output discount90%cheaperTokens coefficient0.2667
gpt-5.6-solOfficial Output$30Our Output$3Output discount90%cheaperTokens coefficient20
gpt-5.6-terraOfficial Output$12Our Output$1.2Output discount90%cheaperTokens coefficient8
gpt-6-astraOfficial Output$50Our Output$5Output discount90%cheaperTokens coefficient33.3333
gpt-6-lunaOfficial Output$0.5Our Output$0.2Output discount60%cheaperTokens coefficient1.3333
gpt-6-solOfficial Output$10Our Output$2Output discount80%cheaperTokens coefficient13.3333
gpt-6.1-solOfficial Output$10Our Output$2Output discount80%cheaperTokens coefficient13.3333
kimi-k2.7-codeOfficial Output$8Our Output$0.8Output discount90%cheaperTokens coefficient5.3333
kimi-k3Official Output$15Our Output$1.5Output discount90%cheaperTokens coefficient10
kimi-k3.1-proOfficial Output$10Our Output$1.5Output discount85%cheaperTokens coefficient10
mimo-v2.6-flashOfficial Output$0.28Our Output$0.05Output discount82.14%cheaperTokens coefficient0.3333
mimo-v2.6-proOfficial Output$0.87Our Output$0.15Output discount82.76%cheaperTokens coefficient1
minimax-m3.1-flashOfficial Output$0.6Our Output$0.1Output discount83.33%cheaperTokens coefficient0.6667
qwen3.8-flashOfficial Output$0.47Our Output$0.1Output discount78.72%cheaperTokens coefficient0.6667
qwen3.8-maxOfficial Output$6Our Output$0.6Output discount90%cheaperTokens coefficient4
Tokens coefficients use up to four decimal places, matching the Models page; trailing zeros are omitted and a positive value below the displayed precision is shown as <0.0001. USD billing uses the exact tariff, not this display value. Missing rates are shown as —, not zero. Discounts are omitted when our price is higher. Your effective account prices are available after sign-in.
The normal table shows synchronous and native async=true pricing. The public default template’s Batch request price coefficient is shown separately above the table, including 1. It applies only to Claude/OpenAI-compatible batch items executed by Model Gate. An existing account can have different effective prices and a different Batch coefficient; check its authenticated Models page.
Your account catalog
The authenticated Models page shows compact rows with Input, Output and Cache read prices in USD per one million provider tokens. The displayed account rate includes the pricing configuration applicable to your account; the official provider reference is informational and is not the amount debited. An official price is struck through only when the account rate is lower. Missing references are displayed as unknown rather than as a zero price.
Search by name, model ID or provider, narrow by provider/capability, or sort by output price. Details opens the model's context limits, advertised capabilities, modalities and the complete five-category rate table, including Cache write and Reasoning. Missing limits remain “Not published.” Prices in the detail table retain their stored decimal precision. Individual accounts additionally see the configured Tokens display coefficients; Business accounts retain USD pricing.
Images pricing is route-specific
A model capability flag and the text-token table do not price image work. Reviewed image contracts can charge per returned image or by disjoint text/image input, cached input and image output usage. Obtain /v1/images/catalog and an uncharged /v1/images/quote for the actual route and account. Unknown usage-dependent price is not zero; accounting_recorded is not confirmation of wallet settlement. See Images quote.
Media tariffs on model cards
Authenticated Models cards also show Images, Audio and Video contract prices, separately from token endpoint rates: all parameter-specific rules, the price formula, the selected account billing multiplier and its source. Depending on the offer, the formula is per image, completed request/video, second, rounded audio minute or normalized token usage. A separate official-media card is for Usage Limit valuation, not an extra wallet charge. The Models token reference table is never substituted for a missing official media tariff.
The card is account-level information, not a request quote: choose your API key and route in Playground for current key/group availability, parameters and Usage multiplier. A missing, expired or unavailable media tariff is not displayed as zero. With official_price valuation, the route must have an official media tariff; absence blocks new work rather than using the customer price.
Separate media lists and supported parameters
The authenticated Models and prices page groups text/other models, image generation/editing, audio and video into separate lists. Classification uses stored media contracts and output modality, not a guess from the model name. A model with more than one media operation can appear in each applicable list; each route keeps its own parameters and tariff.
Media cards show allowed option values and defaults, quantity/duration/size/resolution choices, prompt limits and input/reference/mask limits where present in the reviewed contract. Forbidden parameter combinations are shown separately. A missing or expired contract is explicitly unavailable, not unlimited or free. Global token-price defaults are not displayed as media prices. Details omits the effective token-price section for media models or when no valid token rates exist; a deliberately configured zero token rate remains visible for text models.
Search and sorting stay inside these groups. This account-level view does not select an API key or test supplier health. The public /v1/models endpoint is now a minimal identity list; its disabled extended variant cannot be used to retrieve rates or these parameters.
Tokens coefficients are displayed with up to four decimal places, matching the signed-in Models page. Trailing zeros are omitted; a positive value below display precision appears as <0.0001. Billing continues to use the exact tariff.