Scan verified AI API opportunities, then sort by Deal Score or token pricing and narrow by provider. Checked time shows when the offer was last re-verified.
120Verified
4Freshly checked
0Price drops · 24h
0New · 24h
Deal signals
Deals to notice first
Deal Score remains current-offer attractiveness; it does not replace intrinsic quality.
Price drops0
No verified price drops in the last 24h.
New deals0
No newly recorded deals in the last 24h.
Fresh checks4
4 of 120 verified resources are inside the freshness window.
Cloudflare AI Gateway's current GPT-5.6 Sol model page lists $2.00 per 1M input tokens, $10.00 per 1M output tokens, and $0.25 per 1M cached-input tokens for the openai/gpt-5.6-sol route. The same page still carries an outdated note saying the 50% promotion ended September 18, 2026, so the current pricing is official but the promotion/terms text is internally inconsistent and should be rechecked before treating it as a time-limited deal.
AI Agent ControllerAutomation
Risk: Cloudflare's current model page and historical promotion terms conflict: the current page lists $2 input / $10 output / $0.25 cached per 1M tokens but still displays an expired September 18 promotion note. Do not infer a renewed promotion deadline or guaranteed duration. Unified Billing/BYOK conditions may differ; route performance should not be assumed equal to direct OpenAI.
Microsoft Foundry offers GPT-6 Astra in Standard Global at US$10 per 1M input tokens, US$1 per 1M cached-input tokens, US$12.50 per 1M cache-write tokens, and US$50 per 1M output tokens for short-context requests. The model supports a 1.05M-token context window; requests above 272K input tokens use the higher long-context tariff for the full request.
Anthropic Claude API offers Claude Sonnet 5 at US$2 per 1M input tokens and US$10 per 1M output tokens, with a 1M-token context window and up to 128K output tokens. Anthropic made the US$2/US$10 introductory tariff permanent on August 10, 2026.
Amazon Bedrock offers OpenAI GPT-6 Astra through Global Cross-Region Inference at US$10 per 1M input tokens, US$1 per 1M cache-read tokens, US$12.50 per 1M 30-minute cache-write tokens, and US$50 per 1M output tokens for short-context requests. The model supports a 1.05M-token context window; long-context Global CRIS requests are priced at US$20 input, US$2 cache read, US$25 cache write, and US$75 output per 1M tokens.
Fireworks AI serves GLM 5.3 Flash at $0.15 per 1M input tokens, $0.03 per 1M cached input tokens, and $0.50 per 1M output tokens. The serverless route is Ready with a 1.04M-token context window and image input support.
OpenAI GPT-5.6 Terra costs $2.00 per 1M input tokens, $0.20 per 1M cached input tokens, and $12.00 per 1M output tokens, with a 1.05M-token context window and 128K max output. Requests above 272K input tokens use higher long-context rates.
OpenAI API offers GPT-6 Astra at US$10 per 1M input tokens, US$1 per 1M cached-input tokens, US$12.50 per 1M cache-write tokens, and US$50 per 1M output tokens for standard-context requests. The model supports a 1.05M-token context window and up to 128K output tokens.
OpenAI GPT-5.6 Luna costs $0.20 per 1M input tokens, $0.02 per 1M cached input tokens, and $1.20 per 1M output tokens, with a 1.05M-token context window and 128K max output. Requests above 272K input tokens use higher long-context rates.
Microsoft Foundry offers GPT-5.6 Sol through Standard Global deployments. Azure OpenAI metered API promotional pricing is US$4 per 1M input tokens and US$20 per 1M output tokens from September 1, 2026 through at least November 30, 2026; the published Standard Global table lists cached input at US$0.50 per 1M tokens.
Google Gemini Developer API offers Gemini 3.8 Flash Priority inference at US$1.35 per 1M input tokens, US$0.135 per 1M cached tokens, and US$6.75 per 1M output tokens through December 31, 2026. Priority pricing doubles on January 1, 2027.
Fireworks AI serves DeepSeek V4.1 Flash through its Serverless API at $0.22 per 1M input tokens, $0.007 per 1M cached input tokens, and $0.66 per 1M output tokens. The official model page lists a 1.04M-token context length and marks the model Ready for serverless use.
Risk: Newly listed model with limited production history; low token price should not be interpreted as evidence of model quality, latency, reliability, or throughput.
OpenAI GPT-5.6 Sol currently costs US$4.00 per 1M input tokens, US$0.40 per 1M cached-input tokens, and US$20.00 per 1M output tokens on the native OpenAI API; OpenAI says this promotional pricing is available at least through November 21, 2026. The model has a 1.05M-token context window and 128K maximum output, with higher rates for requests above 272K input tokens.
Google Gemini Developer API offers Gemini 3.8 Flash Flex inference at US$0.375 per 1M input tokens, US$0.0375 per 1M cached tokens, and US$1.875 per 1M output tokens through December 31, 2026. Flex has no free tier and the listed rates double on January 1, 2027.
Fireworks AI serves Thinking Machines Lab's Inkling through its serverless API at $1.00 per 1M input tokens, $0.17 per 1M cached input tokens, and $4.05 per 1M output tokens. Inkling is a 975B Mixture-of-Experts model with 41B active parameters, a 1.04M-token context window, image input, function calling, and serverless availability.
Google Gemini Developer API offers Gemini 3.8 Flash Standard at US$0.75 per 1M input tokens, US$0.075 per 1M cached tokens, and US$3.75 per 1M output tokens through December 31, 2026. The model supports a 1,048,576-token input context and 65,536-token maximum output; standard pricing doubles on January 1, 2027.
Upstage Solar Pro 4 direct API promotional pricing is USD 0.09 per 1M input tokens, USD 0.018 per 1M cached input tokens, and USD 0.36 per 1M output tokens through October 19, 2026 UTC. The official schedule lists USD 0.15/0.03/0.60 starting October 20 UTC; standard displayed list prices are USD 0.30/0.06/1.20 per 1M tokens. The model has a 512K context window.
Risk: Official current promotional schedule: USD 0.09/0.018/0.36 per 1M input/cached/output tokens until 2026-10-20T00:00Z; next scheduled USD 0.15/0.03/0.60 begins October 20 UTC. Displayed standard list price USD 0.30/0.06/1.20 differs from current promotional pricing. Recheck on October 20.
OpenAI GPT-5.6 Cyber costs US$12.50 per 1M input tokens, US$1.25 per 1M cached-input tokens, and US$75 per 1M output tokens on the native OpenAI API. It has a 400K-token context window and requires separate approval and provisioning through OpenAI's Daybreak program for authorized cybersecurity research and testing.
DeepSeek's official deepseek-flash API now serves DeepSeek-V4.1-Flash-1017 with a 1M-token context window. Off-peak rates remain USD $0.15 per 1M uncached input tokens, $0.003 per 1M cached input tokens and $0.60 per 1M output tokens; weekday peak rates are double.
Risk: Peak/off-peak billing means effective cost depends on request timing; legacy aliases are transitional. Provider assertions about benchmark superiority are not used as a Compute Score.
Fireworks AI serves DeepSeek-V4-Flash-Vision-Exp through its Serverless API at $0.22 per 1M input tokens, $0.007 per 1M cached input tokens, and $0.66 per 1M output tokens. The official model page lists a 1.04M-token context length, image input support, Ready status, and identifies the model as experimental.
Risk: Experimental multimodal model with limited production history. Low token pricing and official Ready status are not evidence of durable latency, throughput, output quality, or production reliability.
OpenAI GPT-Image-2.5 Flare is available through the Image API and Responses API image-generation tool. Text input costs US$5/1M tokens (US$1.25/1M cached), image input costs US$8/1M tokens (US$2/1M cached), and image output costs US$30/1M tokens; text output is not billed because the model outputs images.
OpenAI GPT-Image-2.5 Sunburst is available through the Image API and Responses API image-generation tool, targeting higher-precision image generation and editing. Text input costs US$5/1M tokens (US$1.25/1M cached), image input costs US$8/1M tokens (US$2/1M cached), and image output costs US$30/1M tokens; text output is not billed because the model outputs images.
Anthropic Claude API offers Claude Fable 5.1 (claude-fable-5-1) at US$10 per 1M input tokens and US$50 per 1M output tokens, with a 1M-token context window and up to 128K output tokens. Prompt-cache reads cost US$0.25 per 1M tokens, and Batch API input/output pricing is discounted by 50%.
Anthropic offers Claude Mythos 5.1 at US$10 per 1M input tokens and US$50 per 1M output tokens. Access remains limited to vetted users, but Anthropic opened applications for its Life Sciences Verification Program on September 17, 2026, allowing approved organizations to use Mythos through the first-party API console.
Vercel AI Gateway offers Poolside Laguna S 2.1 Free at $0 input and $0 output pricing. The free route uses model ID poolside/laguna-s-2.1-free with a 256K-token context window; Vercel states it serves the same Laguna S 2.1 model as the paid route within the smaller context window.
Vercel AI Gateway currently offers inclusionai/ling-3.0-flash-vl-free at US$0 per 1M input and output tokens via Novita AI, with a 256K-token context window and up to 32K output tokens. The route supports text, image, and video inputs with text output; the current official page does not state a promotional end date.
Cloudflare Workers AI serves @cf/moonshotai/kimi-k2.6 at $0.95 per 1M input tokens, $4.00 per 1M output tokens, and $0.16 per 1M cached input tokens. The model has a 262,144-token context window, supports vision and tool calling, and requires Workers Paid or prepaid AI Gateway credits.
Cloudflare Workers AI serves @cf/zai-org/glm-5.3-flash at $0.15 per 1M input tokens, $0.50 per 1M output tokens, and $0.03 per 1M cached input tokens. The official model page lists a 1,048,576-token context window with function calling, reasoning and vision; Workers Paid or prepaid AI Gateway credits are required.
Cloudflare Workers AI serves @cf/deepseek-ai/deepseek-v4-flash-0731 at $0.44 per 1M input tokens, $1.32 per 1M output tokens, and $0.014 per 1M cached input tokens. The current official model page lists a 1,048,576-token context window; paid access is required via Workers Paid or prepaid AI Gateway credits.
OpenRouter routes TypeSafe Jev 1.13 at $0.042 per 1M input tokens and $0 per 1M output tokens. The model has a 32,000-token context window and returns structured decisions for routing, classification, and other fast decision points.
OpenRouter routes Z.ai GLM 5.3 FlashX at $0.37 per 1M input tokens, $1.25 per 1M output tokens, and $0.075 per 1M cache-read tokens. The route exposes a 1,048,576-token context window and accepts text, image, and video input with tool calling.
Amazon Bedrock offers Moonshot AI Kimi K3 through the US CRIS profile at $3.30 per 1M input tokens and $16.50 per 1M output tokens, with cache reads at $0.33 per 1M tokens. The profile keeps processing within the US geography and supports a 1M-token context window; Standard-tier pricing is recorded here.
Amazon Bedrock offers Moonshot AI Kimi K3 through the Global CRIS profile at $3.00 per 1M input tokens and $15.00 per 1M output tokens, with cache reads at $0.30 per 1M tokens. The model supports a 1M-token context window and cross-Region inference; Standard-tier pricing is recorded here.
OpenAI GPT-Realtime-2.1 Mini is a lower-cost realtime reasoning model for voice interactions. Text pricing is $0.60 per 1M input tokens, $0.06 cached input, and $2.40 per 1M output tokens; audio is $10.00 input and $20.00 output per 1M audio tokens, with a 128,000-token context window.
OpenAI GPT-Realtime-2.1 is a realtime reasoning model for voice-agent workflows. Text pricing is $4.00 per 1M input tokens, $0.40 cached input, and $24.00 per 1M output tokens; audio is $32.00 input and $64.00 output per 1M audio tokens, with a 128,000-token context window.
SpaceXAI Grok 4.7 is available through the public xAI API at $2 per 1M input tokens, $0.50 per 1M cached input tokens, and $6 per 1M output tokens for prompts below 200k tokens. It has a 500,000-token context window; pricing doubles above 200k prompt tokens.
OpenAI API offers GPT-6 Sol at US$2 per 1M input tokens, US$0.20 per 1M cached-input tokens, and US$10 per 1M output tokens for standard short-context requests. It supports a 1.05M-token context window and up to 128K output tokens; long-context, Batch/Flex, Fast, and regional-processing rates differ.
OpenAI API offers GPT-6 Luna at US$0.10 per 1M input tokens, US$0.01 per 1M cached-input tokens, and US$0.50 per 1M output tokens for standard short-context requests. It supports a 1.05M-token context window and up to 128K output tokens; long-context, Batch/Flex, Fast, and regional-processing rates differ.
Anthropic Claude Opus 5.5 costs $4 per 1M input tokens and $20 per 1M output tokens, with cache reads at $0.20 per 1M tokens. Fast mode is also available at $8 input and $40 output per 1M tokens.
DigitalOcean Serverless Inference now serves OpenAI GPT-6 Luna at $0.10 per 1M input tokens and $0.50 per 1M output tokens. The route uses model ID openai-gpt-6-luna with a 1.05M-token context window and supports serverless inference through DigitalOcean's inference API.
DigitalOcean Serverless Inference now serves TypeSafe AI Jev 1.13.0 at $42 per billion input tokens ($0.042 per 1M), with output tokens free. Jev returns typed decisions with calibrated probabilities rather than free-form text and uses the pinned model ID typesafe-jev-1.13.0.
DigitalOcean Serverless Inference now serves OpenAI GPT-6 Sol at $2 per 1M input tokens and $10 per 1M output tokens. The route uses model ID openai-gpt-6-sol with a 1.05M-token context window and supports serverless inference through DigitalOcean's inference API.
Alibaba Cloud Model Studio serves qwen-audio-3.1-realtime-plus in Singapore at $0.80 per 1M text-input tokens, $6.40 per 1M audio-input tokens, $6.40 per 1M text-output tokens, and $24 per 1M audio-output tokens. The real-time duplex model has a 262,144-token context window and official limits of 60 RPM and 100,000 TPM.
Anthropic Claude API offers Claude Opus 4.8 (claude-opus-4-8) at US$5 per 1M input tokens and US$25 per 1M output tokens, with a 1M-token context window and up to 128K output tokens. Prompt-cache reads cost US$0.50 per 1M tokens, and Batch API input/output pricing is discounted by 50%.
OpenAI API offers GPT-6.1 Sol at USD 2 per 1M input tokens, USD 0.10 per 1M cached-input tokens, and USD 10 per 1M output tokens. It has a 1.05M-token context window and 128K max output; long-context and alternate processing modes use different rates.
Amazon Bedrock serves xAI Grok 4.7 through the us.xai.grok-4.7 US geographic cross-Region inference profile at $2.20 per 1M input tokens, $6.60 per 1M output tokens, and $0.55 per 1M cached tokens on the Standard tier. The model has a 500K-token context window and keeps processing within the US geography.
Amazon Bedrock serves xAI Grok 4.7 through the global.xai.grok-4.7 cross-Region inference profile at $2.00 per 1M input tokens, $6.00 per 1M output tokens, and $0.50 per 1M cached tokens on the Standard tier. The model has a 500K-token context window; Global CRIS can route across supported commercial AWS Regions.
Google Gemini 4 Argon (gemini-4-argon) is listed as a stable paid Gemini API model at USD $2 per 1M input tokens, $10 per 1M output tokens and $0.10 per 1M implicitly cached input tokens through November 30, 2026. Standard rates double December 1; its shared context window is 1,048,576 tokens and the Free tier is not available.
Risk: Introductory API prices double on December 1, 2026. A paid tier is required; account-specific quota and successful request are not independently verified.
Cloudflare Workers AI serves @cf/zai-org/glm-5.3 at $1.40 per 1M input tokens, $4.40 per 1M output tokens, and $0.26 per 1M cached input tokens. The model has a 1,048,576-token context window, supports reasoning and function calling, and requires Workers Paid or prepaid AI Gateway credits.
OpenAI API offers GPT-6 Astra in the Ultrafast service tier at USD 60 per 1M short-context input tokens, USD 6 per 1M cached-input tokens, and USD 300 per 1M output tokens. Ultrafast is documented at up to 8× Standard speed, is subject to rate limits, and supports global processing and US data residency but not EU regional processing.
Cloudflare Workers AI serves @cf/moonshotai/kimi-k2.7-code at $0.95 per 1M input tokens, $4.00 per 1M output tokens, and $0.19 per 1M cached input tokens. The model has a 262,144-token context window, supports vision and multi-turn tool calling, and requires Workers Paid or prepaid AI Gateway credits.
Anthropic Claude Sonnet 5.5 (claude-sonnet-5-5) costs $2 per 1M input tokens and $10 per 1M output tokens on the standard first-party Claude API route, with a 1M-token context window. Anthropic reduced its cache-read price from $0.20 to $0.10 per 1M tokens effective October 7, 2026; other token pricing is unchanged.
DeepSeek V4 Pro (DeepSeek-V4-Pro-0813) remains available as deepseek-v4-pro with a 1M-token context window. Off-peak pricing is $0.66/1M cache-miss input, $0.022/1M cache-hit input, and $1.98/1M output; peak pricing is $1.32/$0.044/$3.96.
Risk: V4 Pro is materially more expensive than V4.1 Flash and DeepSeek previously announced a phase-out before reversing that plan; availability should continue to be monitored.
DigitalOcean Serverless Inference offers Anthropic Claude Fable 5.1 at $10 per 1M input tokens, $50 per 1M output tokens, and $0.25 per 1M cache-read tokens. The route uses model ID anthropic-claude-fable-5.1 with a 1M-token context window and supports prompt caching and tool calling.
DigitalOcean Serverless Inference offers OpenAI GPT-6.1 Sol at $2 per 1M input tokens, $10 per 1M output tokens, and $0.10 per 1M cache-read tokens for Standard prompts up to 272K tokens. The route uses model ID openai-gpt-6-1-sol with a 1.05M-token context window; longer prompts use higher rates.
Risk: EU data residency is internally contradictory in current official DigitalOcean documentation (2026-10-09); do not assert compliance or EU routing without account-level verification.
OpenAI gpt-rosalind-research is a specialized life-sciences API model priced at $5.00 per 1M input tokens, $0.50 per 1M cached input tokens, and $25.00 per 1M output tokens. Billing begins October 5, 2026, and access is limited to approved internal research through OpenAI's trusted-access program.
Mistral AI's Mistral Large 4 is a public-preview multimodal Model API with a 1M-token context window. Launch pricing is $0.68 per 1M input tokens, $0.07 per 1M cached input tokens, and $2.09 per 1M output tokens for two weeks at 50% off standard pricing.
Ollama Cloud serves DeepSeek-V4-Pro at $1.32 per million input tokens, $0.044 per million cached input tokens, and $3.96 per million output tokens, with a 1M-token context. Off-peak rates are lower.
Anthropic Claude Haiku 5.5 (claude-haiku-5-5) costs $0.10 per 1M input tokens and $0.50 per 1M output tokens for prompts up to 100,000 tokens; prompts over 100,000 tokens cost $0.50/$2.50 per 1M input/output tokens. The model supports a 1M-token context window and 128K maximum output; cache reads start at $0.01/1M tokens, and Batch API offers 50% off input/output.
Ollama Cloud serves DeepSeek-V4.1-Flash at $0.30 per million input tokens, $0.006 per million cached input tokens, and $1.20 per million output tokens, with a 1M-token context. Off-peak rates are lower.
Ollama Cloud serves GLM-5.3-Flash at $0.15 per million input tokens, $0.03 per million cached input tokens, and $0.50 per million output tokens, with a 1M-token context. The model page lists US and European hosting.
OpenAI GPT-6.1 Sol Ultrafast is available through the Responses API using model gpt-6.1-sol and service_tier=ultrafast. For prompts up to 272,000 input tokens, it costs USD 12 per 1M input tokens, USD 0.60 per 1M cached input tokens, and USD 60 per 1M output tokens; longer prompts cost USD 24/1M input and USD 90/1M output. Ultrafast supports global processing and US/EU data residency, with separate rate limits and regional processing surcharges.
Upstage Solar Mini 4 direct API is USD 0.05 per 1M input tokens, USD 0.005 per 1M cached input tokens, and USD 0.20 per 1M output tokens from October 11 through October 22, 2026 UTC. It has a 512K context window; the official promotion ends October 23 at 00:00 UTC.
Risk: Current October 11–22 UTC promotional phase is USD 0.05/0.005/0.20 per 1M input/cached/output tokens. Scheduled promotional phase ends 2026-10-23T00:00Z; official list rates are USD 0.10/0.01/0.40. Account-specific eligibility/billing remains untested. Compute Score left blank without sufficient independent quality evidence.
DigitalOcean Inference offers Claude Haiku 5.5 (anthropic-claude-5.5-haiku) at USD 0.10 per 1M input tokens and USD 0.50 per 1M output tokens for prompts up to 100K tokens; above 100K, rates are USD 0.50 input and USD 2.50 output per 1M tokens. Cache-read rates are USD 0.01/0.05 per 1M tokens below/above that threshold; prepaid funding is required and EU data residency is unavailable.
Cloudflare Workers AI offers @cf/cloudflare/clef-omni, a 30B-parameter MoE decision model with 3B active parameters, at USD 0.15 per million input tokens and a 64,000-token context window. It accepts text, images, audio, and video and returns typed decision probabilities; output tokens are not billed.
Google Gemini 3.6 Flash Flex API costs $0.375 per 1M input tokens and $1.875 per 1M output tokens through December 31, 2026, with cache reads at $0.0375 per 1M tokens. Flex is a preview, best-effort synchronous tier with variable latency; the model supports a 1,048,576-token input context.
Risk: Flex is Preview and best-effort/sheddable; Google targets roughly 1–15 minute latency and does not guarantee Standard-tier availability.
Microsoft Foundry offers Microsoft-Decision-1 at USD $0.042 per million input tokens with no output-token charge. The text-only decision model scores predefined options instead of generating prose, supports a 32,768-token context, and is available through Microsoft Foundry GlobalStandard and selected DataZoneStandard deployments.
Google Gemini 3.6 Flash Priority API costs $1.35 per 1M input tokens and $6.75 per 1M output tokens through December 31, 2026, with cache reads at $0.135 per 1M tokens. Priority requires eligible Tier 2/3 access, may downgrade overflow to Standard billing, and supports a 1,048,576-token input context.
Risk: Priority tier is limited to eligible Tier 2/3 accounts; over-limit traffic can be downgraded to Standard at Standard rates.
Google Gemini 3.6 Flash standard API costs $0.75 per 1M input tokens and $3.75 per 1M output tokens through December 31, 2026, with cache reads at $0.075 per 1M tokens. It supports a 1,048,576-token input context; the published rates increase on January 1, 2027.
Meta Model API offers muse-wonder-1.0 in public preview at USD $4.00 per 1M input tokens, $0.20 per 1M cached input tokens, and $20.00 per 1M output tokens. The direct API model has a 1,048,576-token context window and supports agentic workflows; Muse Wonder Full is a separate Muse Code subscription-only offering without published direct API token pricing.
OpenRouter lists the Upstage Solar Decide structured-decision route at USD 0.05 per 1M input tokens and USD 0.05 per 1M cached tokens, with no output-token charge, under a displayed 50% discount. The route has a 524,288-token context window and returns typed decisions rather than generated prose; no promotion end date is published on the current model page.
Vercel AI Gateway lists Inception Mercury 2.5 at $0.04 per 1M input tokens, $0.15 per 1M output tokens and $0.004 per 1M cached tokens, with a 260K-token context window. The route is labeled 80% off, with no stated end date.
Upstage Solar Decide is a beta structured-decision API on Upstage Console, billed at USD 0.10 per 1M input tokens and USD 0.10 per 1M cached tokens, with no output-token charge. It uses the System One endpoint to return typed decisions and probabilities rather than generated prose, with a 512K-token context window.
Vercel AI Gateway offers Glyph Cluster (model ID stealth/glyph-cluster) at USD $0 per 1M input and output tokens during its stealth early-access period, with a 256K-token context window. Access requires a Pro or Enterprise team with purchased AI Gateway credits; the model is text-only and does not support Zero Data Retention.
Risk: No Zero Data Retention: prompts and outputs may be retained and used for model training; limited stealth early access, no stated end date; requires paid team plan and purchased credits.
DigitalOcean offers Arcee Trinity Large Thinking in public preview for serverless inference at USD $0.25 per 1 million input tokens, $0.90 per 1 million output tokens, and $0.06 per 1 million cache-read tokens. The model ID is arcee-trinity-large-thinking with a 128,000-token context window and a 32,000-token maximum output.
Google Gemini 4 Argon Priority API costs USD $3.6 per 1M input tokens, $18 per 1M output tokens and $0.18 per 1M implicitly cached input tokens through November 30, 2026. Rates double December 1; this paid route has a shared 1,048,576-token context window.
Liner Model API supports the version-pinned liner-mark-1.2 model with a 1,000,000-token context window. Official token prices are USD $1.00 per 1M input tokens, USD $5.00 per 1M output tokens, and USD $0.10 per 1M cached input tokens; usage is deducted from a prepaid credit balance.
Liner Model API supports the version-pinned liner-mark-1.1 model with a 1,000,000-token context window. Official token prices are USD $1.00 per 1M input tokens, USD $5.00 per 1M output tokens, and USD $0.10 per 1M cached input tokens; usage is deducted from a prepaid credit balance.
Google Gemini 4 Argon Batch API costs USD $1 per 1M input tokens, $5 per 1M output tokens and $0.05 per 1M implicitly cached input tokens through November 30, 2026. Rates double December 1; this paid route has a shared 1,048,576-token context window.
Liner Model API supports the version-pinned liner-mark-1.3 model with a 1,000,000-token context window. Official token prices are USD $1.00 per 1M input tokens, USD $5.00 per 1M output tokens, and USD $0.10 per 1M cached input tokens; usage is deducted from a prepaid credit balance.
Google Gemini 4 Argon Flex API costs USD $1 per 1M input tokens, $5 per 1M output tokens and $0.05 per 1M implicitly cached input tokens through November 30, 2026. Rates double December 1; this paid route has a shared 1,048,576-token context window.
Liner Model API supports the version-pinned liner-mark-1.0 model with a 1,000,000-token context window. Official token prices are USD $1.00 per 1M input tokens, USD $6.00 per 1M output tokens, and USD $0.10 per 1M cached input tokens; usage is deducted from a prepaid credit balance.
DigitalOcean offers the open-weight gpt-oss-20b model through DigitalOcean-hosted serverless inference at USD $0.05 per 1 million input tokens and $0.45 per 1 million output tokens. The documented model ID is openai-gpt-oss-20b, with a 128,000-token context window and 4,096 maximum output tokens.
DigitalOcean Inference lists Claude Sonnet 5.5 (model ID anthropic-claude-sonnet-5.5) at USD $2.00 per 1M input tokens, $10.00 per 1M output tokens, and $0.20 per 1M cache-read tokens. The serverless model supports a 1,000,000-token context window and requires prepaid balance or promotional credits; DigitalOcean currently states that EU data residency is unavailable.
OpenAI GPT-4o mini is listed at $0.15 per 1M input tokens, $0.075 per 1M cached input tokens, and $0.6 per 1M output tokens on the native OpenAI API. OpenAI announced on October 14, 2026 that the gpt-4o-mini-2024-07-18 snapshot is deprecated, with API shutdown scheduled for January 14, 2027; its recommended replacement is gpt-6-luna.
OpenAI GPT-4o is listed at $2.5 per 1M input tokens, $1.25 per 1M cached input tokens, and $10 per 1M output tokens on the native OpenAI API. OpenAI announced on October 14, 2026 that the gpt-4o-2024-08-06 and gpt-4o-2024-11-20 snapshots are deprecated, with API shutdown scheduled for January 14, 2027; its recommended replacement is gpt-6-luna.
Vercel AI Gateway offers Microsoft's MAI-Image-2.6-Flash via Azure with listed input pricing of USD 1.75 per million tokens and output pricing of USD 19 per million tokens. The model supports image generation and editing, and the model page advertises a lowest available image configuration of USD 0.01 per image rather than a guaranteed price for every request.
OpenAI GPT-4.1 is listed at $2 per 1M input tokens, $0.5 per 1M cached input tokens, and $8 per 1M output tokens on the native OpenAI API. OpenAI announced on October 14, 2026 that the gpt-4.1-2025-04-14 snapshot is deprecated, with API shutdown scheduled for January 14, 2027; its recommended replacement is gpt-6-luna.
OpenAI GPT-4.1 mini is listed at $0.4 per 1M input tokens, $0.1 per 1M cached input tokens, and $1.6 per 1M output tokens on the native OpenAI API. OpenAI announced on October 14, 2026 that the gpt-4.1-mini-2025-04-14 snapshot is deprecated, with API shutdown scheduled for January 14, 2027; its recommended replacement is gpt-6-luna.
Vercel AI Gateway offers Microsoft's MAI-Image-2.6 via Azure with listed input pricing of USD 5 per million tokens and output pricing of USD 38 per million tokens. The model supports text-to-image generation and image editing; the model page also advertises a lowest available image configuration of USD 0.01 per image, which is not a fixed all-configurations price.
Vercel AI Gateway routes Google Gemini 4 Argon through Google or Google Vertex AI. Listed base pricing starts at $2 per 1M input tokens and $10 per 1M output tokens, with a 1,048,576-token context window; caching, grounding, and higher usage tiers may have separate charges.
Vercel AI Gateway provides Meta Muse Wonder 1.0 via the Meta route at $4 per 1M input tokens, $20 per 1M output tokens, and $0.20 per 1M cached input tokens. The model lists a 1M-token context window and supports long-running agent and computer-use workloads.
OpenRouter offers the Apodex 1.1 Mini model via apodex/apodex-1.1-mini:free at USD 0 per 1M input tokens and USD 0 per 1M output tokens, with a 262,144-token context window. Free model requests are rate limited to 20 per minute and 50 per day for accounts without qualifying purchased credits.
Risk: Free route is rate-limited and availability may vary; no SLA or sustained throughput guarantee established.
OpenRouter lists Perplexity Decider V1.1 27B at USD 0.02 per 1M input tokens and USD 0 per 1M output tokens. The 262,144-token route uses OpenRouter's Decisions API to return typed probabilistic decisions from text or image inputs rather than chat text, with up to 128 questions per request.
OpenRouter lists NVIDIA Nemotron 3.5 Super VL (free route) at USD $0 per 1M input and output tokens, with a 1,000,000-token context window for multimodal text, image, and video input. The free NVIDIA endpoint prohibits confidential or personal data and may log usage; rate limits and continued free availability are not guaranteed.
OpenRouter offers Unbiased Pareto 26.9 at USD 2.50 per 1M input tokens, USD 7.50 per 1M output tokens, and USD 0.25 per 1M cached input tokens. The versioned multimodal route supports a 262,144-token context window, up to 131,072 output tokens, text/image input and tool calling.
OpenRouter lists TypeSafe Jev 1.14 at USD 0.042 per 1 million input tokens and USD 0 per 1 million output tokens, with a 262,144-token context window. This specialized model returns structured decisions for routing and classification rather than general-purpose prose.
Risk: Specialized structured-decision API; not directly comparable to general text-generation models. Route and provider terms may change.
OpenRouter lists Liquid AI LFM2.5-24B-A2B at USD 0.10 per 1M input tokens, USD 1.00 per 1M output tokens, and USD 0.05 per 1M cached input tokens. The route has a 128,000-token context window, an 8,192-token output limit, and supports tool calling and JSON-schema structured outputs.
DigitalOcean Inference offers NVIDIA Nemotron 3.5 Super VL (model ID nemotron-3.5-super-vl) for serverless inference at $0.30 per 1M input tokens, $0.65 per 1M output tokens, and $0.06 per 1M cached input tokens. It supports text, image, document and GUI screenshot inputs with a 1,048,576-token context window; audio input is not supported.
OpenRouter lists NVIDIA Nemotron 3.5 Super VL Standard routing from USD $0.10 per 1M input tokens and $0.40 per 1M output tokens through its lowest-priced listed provider, with a 1,048,576-token context window. Other provider routes cost more, so the displayed price is a routed starting price rather than a fixed universal tariff.
Together AI offers Meta Muse Spark Open on its paid serverless API at USD $1.25 per 1M input tokens, $0.15 per 1M cached input tokens and $4.25 per 1M output tokens. The published model identifier is meta-models/Muse-Spark-Open-816B-A42B, with an advertised 1M-token context window.
Modal Auto Endpoints offers Meta Muse Spark Open (816B total, 42B active parameters) as a live OpenAI-compatible shared endpoint at USD $1.25 per 1M input tokens, $0.15 per 1M cached input tokens, and $4.25 per 1M output tokens. The model supports a 1M-token context window and text, image, and video inputs; billing is per token.
Risk: Shared Endpoints billed from first token; Modal Starter included compute credits do not cover Shared Endpoint token usage. Latency, throughput and per-account limits not independently tested.
Cloudflare Workers AI offers @cf/daily-co/phonellm-mini-alpha-2 for USD 0.06 per million input tokens and USD 0.24 per million output tokens, with a 262,140-token context window. The Cloudflare-hosted model supports function calling and is tuned for realtime voice-agent workloads.
Fireworks AI lists Meta Muse Spark Open serverless at USD $1.25 per 1M input tokens, $0.15 per 1M cached input tokens and $4.25 per 1M output tokens. The model has a 1,040k-token context window; serverless availability is a US-only limited preview through October 29, 2026, with the usual 50% US-only premium waived.
OpenRouter lists Meta Muse Spark Open at USD $1.25 per 1M input tokens and $4.25 per 1M output tokens, with a 1,048,576-token context window. Its API routes to multiple upstream providers including Together, Nebius Token Factory, Fireworks and Baseten; actual availability and caching terms depend on routing.
Vercel AI Gateway offers perplexity/pplx-decider-flash-v1.2, a 4B multimodal decision model for high-volume classification, routing and filtering, at USD 0.01 per 1 million input tokens. The model supports a 262,144-token context window; the official model page lists input-only pricing, with no output-token rate published.
Vercel AI Gateway currently lists Reflection Beam 1.0 free at $0 per million input and output tokens, with a stated 1M-token context window. The free offer is temporary and the dedicated free route stops serving when the offer ends.
OpenRouter lists XingChen Xing4.0 29B A4B on the free API route at USD 0 per 1M input and output tokens, with a 262,144-token context window. The free endpoint is rate-limited, currently served through Phala, and is not available through EU in-region routing; availability and rate limits are not guaranteed.
Risk: Free endpoints rate-limited; upstream availability variable; no EU in-region routing. Not a guaranteed long-term free entitlement or account-received credit.
Vercel AI Gateway offers perplexity/pplx-decider-v1.2, a 27B multimodal decision model for classification, routing and scoring text, structured records and images, at USD 0.02 per 1 million input tokens. The model supports a 262,144-token context window; the official model page lists input-only pricing, with no output-token rate published.
Vercel AI Gateway offers Elorian Glimpse (elorian/glimpse) at $0.60 per million input tokens and $3.00 per million output tokens, with cached reads at $0.06 per million. The visual reasoning model has a 220,000-token context window and a 4,096-token maximum output, and supports image and video workflows.
OpenRouter lists Reflection Beam 1.0 501B A23B at a starting routed price of USD 0.25 per 1M input tokens and USD 1.00 per 1M output tokens, with a 1,048,576-token context window. This is a dynamic provider-routing floor, currently offered through Nebius Token Factory; other upstream providers have higher rates and EU in-region routing is unavailable.
Risk: Multi-provider route prices and performance vary; advertised floor may not be available for each request. EU in-region routing unavailable. No account-specific API invocation tested.
Vercel AI Gateway offers liquid/d1-ultra, a multimodal decision model for classification, routing and scoring text and images, at USD 0.04 per 1 million input tokens. The model supports a 32,768-token context window; the official model page lists input-only pricing, with no output-token rate published.
Vercel AI Gateway lists DeepSeek V4.1 Flash 1017 at $0.15 per million input tokens and $0.60 per million output tokens with a stated 1M-token context window.
Together AI offers Reflection Beam 1.0 501B A23B through its paid serverless API at USD 0.40 per 1M input tokens, USD 1.20 per 1M output tokens and USD 0.04 per 1M cached input tokens. The official October 16, 2026 launch lists a 1,048,576-token context window, NVFP4 quantization, function calling and structured outputs.
Together AI serves NVIDIA Nemotron 3.5 Super VL on its paid serverless API at USD 0.15 per 1M input tokens, USD 0.50 per 1M output tokens and USD 0.02 per 1M cached input tokens. The October 15, 2026 official release lists a 262,144-token context, NVFP4 quantization, image input, function calling and structured outputs.
OpenRouter lists Unbiased Pareto 26.10 (unbiased/pareto-26.10) at USD 0.80 per 1 million input tokens, USD 3.20 per 1 million output tokens, and USD 0.03 per 1 million cached input tokens. It supports a 1,048,576-token context window and tool calling; EU in-region routing is not currently available.
Risk: No EU in-region routing on the official model page; dynamic routed availability and model performance are not independently benchmarked.
Mistral Small 4 (mistral-small-2603) is a GA model with a 256k-token context window. Standard direct Mistral API pricing is $0.15 per 1M input tokens, $0.015 per 1M cached input tokens, and $0.60 per 1M output tokens.
Cloudflare Workers AI offers @cf/cloudflare/clairic in preview for $0.10 per 1 million input tokens, with no output-token charge. The model supports a 65,536-token context window and returns typed extracted, computed, estimated or forecast values; preview limitations apply.
Risk: Preview; output confidence is not a correctness guarantee; no SLA or availability by account independently verified.
Mistral Large 3 (mistral-large-2512) is a GA multimodal Model API with a 256k-token context window. Standard direct Mistral API pricing is $0.50 per 1M input tokens, $0.05 per 1M cached input tokens, and $1.50 per 1M output tokens.
Mistral Medium 3.5 is a GA multimodal API model (v26.04) with a 256k-token context window. Mistral lists standard API pricing of $1.50 per 1M input tokens, $0.15 per 1M cached input tokens, and $7.50 per 1M output tokens.
Cloudflare Workers AI hosts @cf/google/gemma-4-26b-a4b-it at $0.10 per 1M input tokens and $0.30 per 1M output tokens, with a 256,000-token context window. The official model card lists vision, reasoning, function calling and batch support; any applicable Workers AI plan limits and paid usage charges are separate.
Gemini 3.8 Live publishes audio input/output minute rates, while OpenAI splits voice pricing across GPT-Live-1, realtime models, transcription and translation. Compare the billing layers before estimating agent cost.