Inference offers, sourced and dated.

19 offers from 13 providers, each quoted from the provider’s own page with the date it was checked. Listed by provider, not ranked: a rate only means something against the work you run.

Reference data from third parties. AgenticSensei itself is not live yet.

The same records for agents, as JSON Email me when assessments open

Rates
US-dollar list rates per 1M tokens, as the source page printed them on the checked date.
Not stated
The source page did not say. It is not zero and not a default.
Review by
2026-10-20. After that date, treat every record as stale and check the source.
Whose offers
The providers’, not ours. Check the source before acting on any of them.

Google

Gemini API free tier

Access
Free tier
Cost
No charge
Commercial-use conditions
  • Grounding with Google Search and Grounding with Google Maps are not available on the free tier
Data use
  • Content used to improve our products
Source
ai.google.dev, checked 2026-09-20

Models: gemini-3.8-flash, gemini-3.7-flash, gemini-3.6-flash, gemini-3.5-flash, gemini-3.5-flash-lite

Not stated by the source: Account required, Payment method on file, Phone verification, Organisation required, Regions, Waitlist, Rate limits, Context limits, Conditions, Commercial use, Expiry.

Gemini API paid tier, 3.x Flash promotional rate through 2026-12-31

Access
Paid, metered
Cost
Metered, per model, below
Conditions
  • promotional rate stated as running through December 31, 2026
Data use
  • Content not used to improve our products
Expires
2026-12-31
Source
ai.google.dev, checked 2026-09-20
ModelInput, per 1M tokensOutput, per 1M tokens
gemini-3.8-flash$0.75$3.75
gemini-3.7-flash$0.75$3.75
gemini-3.6-flash$0.75$3.75

Not stated by the source: Account required, Payment method on file, Phone verification, Organisation required, Regions, Waitlist, Rate limits, Context limits, Commercial use, Commercial-use conditions.

Gemini API paid tier, 3.5 Flash and Flash-Lite

Access
Paid, metered
Cost
Metered, per model, below
Data use
  • Content not used to improve our products
Source
ai.google.dev, checked 2026-09-20
ModelInput, per 1M tokensOutput, per 1M tokens
gemini-3.5-flash$1.50$9.00
gemini-3.5-flash-lite$0.30$2.50

Not stated by the source: Account required, Payment method on file, Phone verification, Organisation required, Regions, Waitlist, Rate limits, Context limits, Conditions, Commercial use, Commercial-use conditions, Expiry.

Gemma Terms of Use open weights

Access
Open weights, self-hosted
Cost
Self-hosted: no per-token rate; you run the weights on your own compute
Commercial use
permitted with conditions
Commercial-use conditions
  • the use restrictions must be included as an enforceable provision in any agreement with third party recipients
  • distributions other than through a Hosted Service must be accompanied by a Notice text file
  • modified files must carry prominent notices stating that you modified the files
  • Google reserves the right to restrict (remotely or otherwise) usage it reasonably believes violates the agreement
  • use is prohibited for anything in the Gemma Prohibited Use Policy
Source
ai.google.dev, checked 2026-09-20

Not stated by the source: Account required, Payment method on file, Phone verification, Organisation required, Regions, Waitlist, Rate limits, Context limits, Conditions, Data use, Expiry, Models.

OpenAI

OpenAI API standard processing rates

Access
Paid, metered
Cost
Metered, per model, below
Conditions
  • these are the Standard, Short context rates; the same table prints separate Long context rates, and the Batch table prints lower ones
Source
developers.openai.com, checked 2026-09-20
ModelInput, per 1M tokensOutput, per 1M tokens
gpt-6-astra$10.00$50.00
gpt-5.6-sol$4.00$20.00
gpt-5.6-terra$2.00$12.00
gpt-5.6-luna$0.20$1.20
gpt-4o$2.50$10.00
gpt-4o-mini$0.15$0.60

Not stated by the source: Account required, Payment method on file, Phone verification, Organisation required, Regions, Waitlist, Rate limits, Context limits, Commercial use, Commercial-use conditions, Data use, Expiry.

Anthropic

Claude API base input and output rates

Access
Paid, metered
Cost
Metered, per model, below
Context limits
  • Claude 4.6 and later models include the full 1M token context window at standard pricing, which covers Claude Fable 5.1, Claude Opus 5, Claude Sonnet 5 and Claude Sonnet 4.6 in this offer
  • the cited page does not state a context window for Claude Haiku 4.5
Source
platform.claude.com, checked 2026-09-20
ModelInput, per 1M tokensOutput, per 1M tokens
claude-fable-5-1$10.00$50.00
claude-opus-5$5.00$25.00
claude-sonnet-5$2.00$10.00
claude-sonnet-4-6$3.00$15.00
claude-haiku-4-5$1.00$5.00

Not stated by the source: Account required, Payment method on file, Phone verification, Organisation required, Regions, Waitlist, Rate limits, Conditions, Commercial use, Commercial-use conditions, Data use, Expiry.

Groq

GroqCloud Free plan rate limits

Access
Free tier
Cost
No charge
Rate limits
  • canopylabs/orpheus-arabic-saudi and canopylabs/orpheus-v1-english: 10 RPM, 100 RPD, 1.2K TPM, 3.6K TPD
  • meta-llama/llama-prompt-guard-2-22m and -2-86m: 30 RPM, 14.4K RPD, 15K TPM, 500K TPD
  • openai/gpt-oss-120b, openai/gpt-oss-20b and openai/gpt-oss-safeguard-20b: 30 RPM, 1K RPD, 8K TPM, 200K TPD
  • qwen/qwen3.8-27b: 30 RPM, 1K RPD, 8K TPM, 200K TPD
  • whisper-large-v3 and whisper-large-v3-turbo: 20 RPM, 2K RPD, 7.2K ASH, 28.8K ASD
Conditions
  • the table on this page shows Free Plan limits beside Developer Plan limits: "Upgrade to Developer plan to access higher limits, Batch and Flex processing, and more"
Source
console.groq.com, checked 2026-09-20

Models: canopylabs/orpheus-arabic-saudi, canopylabs/orpheus-v1-english, meta-llama/llama-prompt-guard-2-22m, meta-llama/llama-prompt-guard-2-86m, openai/gpt-oss-120b, openai/gpt-oss-20b, openai/gpt-oss-safeguard-20b, qwen/qwen3.8-27b, whisper-large-v3, whisper-large-v3-turbo

Not stated by the source: Account required, Payment method on file, Phone verification, Organisation required, Regions, Waitlist, Context limits, Commercial use, Commercial-use conditions, Data use, Expiry.

GroqCloud serverless per-token rates

Access
Paid, metered
Cost
Metered, per model, below
Context limits
  • openai/gpt-oss-120b: 131,072 tokens
  • openai/gpt-oss-20b: 131,072 tokens
  • qwen/qwen3.8-27b: 131,042 tokens
Conditions
  • qwen/qwen3.8-27b is a preview model: "Preview models are intended for evaluation purposes only and should not be used in production environments as they may be discontinued at short notice"
Source
console.groq.com, checked 2026-09-20
ModelInput, per 1M tokensOutput, per 1M tokens
openai/gpt-oss-120b$0.15$0.60
openai/gpt-oss-20b$0.075$0.30
qwen/qwen3.8-27b$0.80$4.00

Not stated by the source: Account required, Payment method on file, Phone verification, Organisation required, Regions, Waitlist, Rate limits, Commercial use, Commercial-use conditions, Data use, Expiry.

Cloudflare

Workers AI free daily Neuron allocation

Access
Free tier
Cost
No charge
Rate limits
  • 10,000 Neurons per day at no charge on both the Workers Free and Workers Paid plans
Conditions
  • Workers Free accounts are not eligible for paid usage beyond the free allocation
  • some models require a paid billing method
Source
developers.cloudflare.com, checked 2026-09-20

Not stated by the source: Account required, Payment method on file, Phone verification, Organisation required, Regions, Waitlist, Context limits, Commercial use, Commercial-use conditions, Data use, Expiry, Models.

Workers AI paid per-token rates

Access
Paid, metered
Cost
Metered, per model, below
Conditions
  • $0.011 per 1,000 Neurons above the free allocation
Source
developers.cloudflare.com, checked 2026-09-20
ModelInput, per 1M tokensOutput, per 1M tokens
@cf/meta/llama-3.2-1b-instruct$0.027$0.201
@cf/meta/llama-3.1-70b-instruct-fp8-fast$0.293$2.253
@cf/mistral/mistral-7b-instruct-v0.1$0.11$0.19

Not stated by the source: Account required, Payment method on file, Phone verification, Organisation required, Regions, Waitlist, Rate limits, Context limits, Commercial use, Commercial-use conditions, Data use, Expiry.

OpenRouter

OpenRouter :free model variants

Access
Free tier
Cost
No charge
Rate limits
  • 20 requests per minute
  • the table states 50 requests per day below 10 credits purchased all-time and 1,000 requests per day at 10 or more
  • the text below that table states the higher daily ceiling is granted one credit below the table threshold, so 1,000 requests per day applies from 9 credits
Conditions
  • the limits above apply to model IDs ending in :free; the numbers are rendered client-side from page constants, so scripts/refresh-inference-offers.mjs cannot read them from the served HTML
Source
openrouter.ai, checked 2026-09-20

Not stated by the source: Account required, Payment method on file, Phone verification, Organisation required, Regions, Waitlist, Context limits, Commercial use, Commercial-use conditions, Data use, Expiry, Models.

Hugging Face

Inference Providers monthly credits for free users

Access
Trial credit
Cost
No charge within a credit of $0.10 per month
Account required
Yes
Conditions
  • Free users receive $0.10 of monthly credits, subject to change
  • credits apply only to requests routed by Hugging Face, not to a custom provider key
  • continuing past the monthly credits requires purchasing credits
Source
huggingface.co, checked 2026-09-20

Not stated by the source: Payment method on file, Phone verification, Organisation required, Regions, Waitlist, Rate limits, Context limits, Commercial use, Commercial-use conditions, Data use, Expiry, Models.

Mistral AI

Mistral Free plan monthly API credits

Access
Trial credit
Cost
No charge within a credit of $10.00 per month
Rate limits
  • Limited messages and web searches
  • Limited coding sessions
Source
mistral.ai, checked 2026-09-20

Not stated by the source: Account required, Payment method on file, Phone verification, Organisation required, Regions, Waitlist, Context limits, Conditions, Commercial use, Commercial-use conditions, Data use, Expiry, Models.

Mistral API per-token rates

Access
Paid, metered
Cost
Metered, per model, below
Source
mistral.ai, checked 2026-09-20
ModelInput, per 1M tokensOutput, per 1M tokens
Mistral Medium 3.5$1.50$7.50
Mistral Small 4$0.15$0.60
Mistral Large 3$0.50$1.50
Ministral 3 (3B)$0.10$0.10
Ministral 3 (8B)$0.15$0.15
Ministral 3 (14B)$0.20$0.20

Not stated by the source: Account required, Payment method on file, Phone verification, Organisation required, Regions, Waitlist, Rate limits, Context limits, Conditions, Commercial use, Commercial-use conditions, Data use, Expiry.

Cerebras

Cerebras Inference Free Trial tier

Access
Free tier
Cost
No charge
Rate limits
  • gpt-oss-120b and qwen-3.8-27b: 5 requests per minute
  • 30K uncached tokens per minute and 90K total tokens per minute
  • 1M tokens per hour and 1M tokens per day
  • qwen-3.8-27b: 2 images per request, 10 MiB maximum payload
  • Rate limits apply at the organization level, not the user level
Source
inference-docs.cerebras.ai, checked 2026-09-20

Models: gpt-oss-120b, qwen-3.8-27b

Not stated by the source: Account required, Payment method on file, Phone verification, Organisation required, Regions, Waitlist, Context limits, Conditions, Commercial use, Commercial-use conditions, Data use, Expiry.

DeepSeek

DeepSeek API peak-hours cache-miss rates

Access
Paid, metered
Cost
Metered, per model, below
Rate limits
  • deepseek-flash: concurrency limit 2500
  • deepseek-v4-pro: concurrency limit 500
Context limits
  • context length 1M
  • max output maximum 384K
Conditions
  • the prices recorded here are the peak, cache-miss rates; off-peak rates are half of the peak rates
  • peak hours are 01:00-04:00 and 06:00-10:00 UTC, Monday through Friday, excluding Chinese public holidays
Source
api-docs.deepseek.com, checked 2026-09-20
ModelInput, per 1M tokensOutput, per 1M tokens
deepseek-flash$0.30$1.20
deepseek-v4-pro$1.32$3.96

Not stated by the source: Account required, Payment method on file, Phone verification, Organisation required, Regions, Waitlist, Commercial use, Commercial-use conditions, Data use, Expiry.

Together AI

Together AI serverless per-token rates

Access
Paid, metered
Cost
Metered, per model, below
Source
www.together.ai, checked 2026-09-20
ModelInput, per 1M tokensOutput, per 1M tokens
MiniMax M3$0.30$1.20
Llama 3.3 70B$1.04$1.04
Qwen3.5 9B$0.17$0.25
GLM-5.3-Flash$0.15$0.50
DeepSeek V4.1 Flash$0.30$1.20

Not stated by the source: Account required, Payment method on file, Phone verification, Organisation required, Regions, Waitlist, Rate limits, Context limits, Conditions, Commercial use, Commercial-use conditions, Data use, Expiry.

Meta

Llama 4 Community License open weights

Access
Open weights, self-hosted
Cost
Self-hosted: no per-token rate; you run the weights on your own compute
Commercial use
permitted with conditions
Commercial-use conditions
  • if, on the Llama 4 version release date, the licensee’s products or services have greater than 700 million monthly active users in the preceding calendar month, the licensee must request a licence from Meta, which Meta may grant at its sole discretion, and is not authorized to exercise any rights under the agreement unless or until Meta expressly grants them
  • distributing or making available the Llama Materials, or a product or service containing them, requires providing a copy of the agreement and prominently displaying "Built with Llama" on a related website, user interface, blogpost, about page, or product documentation
  • using the Llama Materials or their outputs or results to create, train, fine tune or otherwise improve an AI model that is distributed or made available requires including "Llama" at the beginning of that model’s name
Source
developer.meta.com, checked 2026-09-20

Models: Llama 4

Not stated by the source: Account required, Payment method on file, Phone verification, Organisation required, Regions, Waitlist, Rate limits, Context limits, Conditions, Data use, Expiry.

Alibaba (Qwen)

Qwen3-8B weights under Apache-2.0

Access
Open weights, self-hosted
Cost
Self-hosted: no per-token rate; you run the weights on your own compute
Context limits
  • 32,768 token native context length
  • 131,072 tokens with YaRN scaling
Source
huggingface.co, checked 2026-09-20

Models: Qwen/Qwen3-8B

Not stated by the source: Account required, Payment method on file, Phone verification, Organisation required, Regions, Waitlist, Rate limits, Conditions, Commercial use, Commercial-use conditions, Data use, Expiry.