Skip to main content
Models & API pricing

Know what is available—and what each request costs.

One SamsarOne credit balance covers generation and managed inference. Compare model keys first, then expand only the endpoint group you need.

Compare modelsBrowse endpoint rates
Aligned with the current Samsar model and billing registries.
100 creditsequals $1 USD

Usage-priced routes convert provider cost into credits before applying the documented multiplier.

20–60 credits/secexpress video

The selected video model and final billable duration determine the base render charge.

1 credit/secoptional express CTA

Narrator avatar generation is a separate optional stage at 4 credits/sec.

Availability

Model keys by setting

These are request and configuration keys—not universal guarantees. Runtime availability depends on the providers connected to a deployment. Check the live model catalog →

Browse model keys5 configurable settings
Express imageimage_model
GPTIMAGE2NANOBANANAPROSEEDREAMWAN2.7PRO
Express videovideo_model
COSMOS3SUPERI2VRUNWAYMLSEEDANCEI2VSEEDANCE2.5I2VVEO3.1I2VFASTKLINGIMGTOVID3PROKLINGIMGTOVIDTURBOHAPPYHORSEI2VVEO3.1I2V
Planning & assistantinference_model
gpt-5.6-solgemini-3.1-proQWEN3.8kimi-k3
Backing trackbackingtrack_model
LYRIA3ELEVENLABS_MUSIC
Text to speechtts_model
OPENAIELEVENLABSGOOGLE
Express video

Compare video model rates

Express text-to-video and image-list-to-video requests bill the selected rate for each final rendered second. Image-list requests default to RUNWAYML when video_model is omitted.

Video model comparison9 models · credits per second
ModelRequest keyDurationsAspect ratiosCredits/sec
Nvidia Cosmos 3COSMOS3SUPERI2V5 or 7.875 sec at 24 fps16:9 · 9:1620
RunwayMLRUNWAYML5 · 10 sec16:9 · 9:1630
Seedance 1.5SEEDANCEI2V5 · 10 sec16:9 · 9:1630
Seedance 2.5 I2VSEEDANCE2.5I2V5 · 10 · 15 sec16:9 · 9:1650
Veo 3.1 Fast I2VVEO3.1I2VFAST6 · 8 sec16:9 · 9:1636
Kling 3 Pro I2VKLINGIMGTOVID3PRO5 · 10 sec16:9 · 9:1636
Kling 3 Turbo Img2VidKLINGIMGTOVIDTURBO5 · 10 sec16:9 · 9:1636
Happy Horse 1.1 I2VHAPPYHORSEI2V5 · 10 · 15 sec16:9 · 9:1636
Veo 3.1 I2VVEO3.1I2V6 · 8 sec16:9 · 9:1660
Optional stagesadd_narrator_avatar adds 4 credits/sec.express_cta_generation adds 1 credit/sec.See the billing formulas →
Express image

Image model registry rates

Values below are model-registry base credits per generated image. They are not added on top of the express video per-second rates, and the public /image/text_to_image route has its own flat rate.

Image model comparison4 models · credits per image
ModelRequest key1:116:99:16
GPT Image 2GPTIMAGE2464646
Nano Banana ProNANOBANANAPRO151515
SeedreamSEEDREAM152323
Wan 2.7 ProWAN2.7PRO151515
Managed inference

Token pricing without the prose wall

Rates are credits per one million tokens after the route multiplier. QWEN3.8 uses Qwen 3.8 Max for both text and vision across supported adapters; billing follows the provider model returned with actual usage.

Managed inference rate tablesExternal chat 1.25× · assistant 1.5×
External chat

1.25× managed-inference rate

API details →
Request keyAdapterContextInputCachedOutput
gpt-5.6-solStandard≤ 272k input62562.53,750
gpt-5.6-solLong context> 272k input1,2501255,625
gemini-3.1-proStandard≤ 200k input250251,500
gemini-3.1-proLong context> 200k input500502,250
QWEN3.8Qwen 3.8 MaxStandard312.5312.5937.5
QWEN3.8Qwen 3.8 Max≤ 256k input5050200
QWEN3.8Qwen 3.8 Max · long context> 256k input150150600
kimi-k3StandardStandard37537.51,875
Session assistant

1.5× assistant rate

API details →
Request keyAdapterContextInputCachedOutput
gpt-5.6-solStandard≤ 272k input750754,500
gpt-5.6-solLong context> 272k input1,5001506,750
gemini-3.1-proStandard≤ 200k input300301,800
gemini-3.1-proLong context> 200k input600602,700
QWEN3.8Qwen 3.8 MaxStandard3753751,125
QWEN3.8Qwen 3.8 Max≤ 256k input6060240
QWEN3.8Qwen 3.8 Max · long context> 256k input180180720
kimi-k3StandardStandard450452,250
API rate card

Endpoint pricing, grouped by job

Open a group for the billing unit and route-level rate. Full request and response contracts stay in the API reference.

VideoGeneration, translation, edits, and render controls9 rates
POST/video/text_to_video
Rate: 20–60 credits/sec
Billed by: Selected express model × final duration
POST/video/image_list_to_video
Rate: 20–60 credits/sec
Billed by: Same model rates; defaults to RUNWAYML
add_narrator_avatar
Rate: +4 credits/sec
Billed by: Optional image-list stage
express_cta_generation
Rate: +1 credit/sec
Billed by: Optional image-list stage
POST/video/translate_video
Rate: 3 credits/sec
Billed by: Source session duration
POST/video/join_videos
Rate: 3 credits/sec
Billed by: Combined source duration
Outro and footer image updates
Rate: 75 credits/request
Billed by: Add or update routes, including external-user aliases
POST/video/reroll-layers
Rate: Variable
Billed by: Selected image requests + selected AI-video stages
Clone, subtitles, cancel
Rate: No additional charge
Billed by: Current clone and render-control routes
ImageEnhancement, editing, receipt extraction, and rollups8 rates
POST/image/text_to_image
Rate: 15 credits/output
Billed by: Public image generation route
POST/image/enhance
Rate: 11 / 15 / 22 / 29 credits
Billed by: 0.5k / 1k / 2k / 4k output
POST/image/remove_branding
Rate: 15 credits/request
Billed by: Standard image-edit rate
POST/image/add_image_set
Rate: 15 credits/image
Billed by: Requested output count
Receipt template create / read
Rate: No charge
Billed by: Template setup and structured JSON reads
POST/image/receipt_templates/query
Rate: 50 credits/request
Billed by: Validate and extract one receipt
POST/image/create_rollup_banner
Rate: 30 credits/request
Billed by: One rollup render
Rollup with enhance
Rate: Enhance total + 30 credits
Billed by: Only undersized inputs are enhanced
Inference & dataChat, assistants, narratives, moderation, and embeddings9 rates
POST/chat/enhance
Rate: 30 credits/request
Billed by: Synchronous copy enhancement
POST/image/assign_title
Rate: Usage based · 1.5×
Billed by: Actual inference usage
External chat completion POST aliases
Rate: Usage based · 1.25×
Billed by: Token rates in the external inference table
/assistant/completion
Rate: Usage based · 1.5×
Billed by: Token rates in the assistant table
Narrative create_single / create_branching
Rate: Usage based · 1.5×
Billed by: Actual inference usage
Standard embedding routes
Rate: 100 credits / 1M tokens
Billed by: Token volume
Plain-text embedding route
Rate: 250 credits / 1M tokens
Billed by: 2.5× ingestion multiplier
URL embedding routes
Rate: 250 credits / 1M tokens + crawl
Billed by: 2.25 credits per Firecrawl credit
External moderation
Rate: No charge
Billed by: API-key authenticated provider wrapper
Audio, delivery & utilitiesHosted audio, publications, account controls, and status6 rates
External text to speech
Rate: 1 credit/request
Billed by: Hosted external audio route
External sound effect
Rate: 1 credit/request
Billed by: Hosted external audio route
External text to music
Rate: 2 credits/request
Billed by: Hosted external audio route
Publication CRUD and archive
Rate: No additional charge
Billed by: Publish, edit, revoke, and archive controls
Interactive publication metadata
Rate: Usage based · 1.5×
Billed by: Actual inference usage
Status and model discovery
Rate: No charge
Billed by: Polling and runtime availability reads
Final billing

Use response headers as the receipt.

Estimates are useful for validation, but final charges can depend on actual duration, token usage, selected adapters, and optional stages.

x-credits-chargedx-credits-remaining