Your cookie choices

We use necessary cookies to make this site work. With your permission, analytics cookies help us understand usage and marketing cookies support advertising.

You can reject non-essential cookies and still use the site. Change your choices anytime in Cookie settings.

Cookie settingsCookie policy

Featherless AI review

Updated September 28, 2026 · DeployCue Editorial Team · Facts checked September 26, 2026

We earn a commission when you sign up through links on this site. Partner status can affect where a brand appears in our rankings. Partners with a live offer in your country come first, then every other brand by score.

Our verdict

Featherless AI suits developers and small teams who want a long list of open weight models behind an OpenAI-compatible endpoint. Its catalogue spans DeepSeek 4, DeepSeek 3.2 and GLM 5.3, and the largest context window reaches 262,144, enough for long documents and code.

Key facts

Best for
Many open weight models behind an OpenAI-compatible endpoint on a tight budget
Input price from
$0.010 to $3↗
Output price from
$0.020 to $15↗
Models offered
DeepSeek 4, DeepSeek 3.2, GLM 5.3, Kimi 3, Qwen 3.8, Qwen 3.6, Gemma 4, Mistral 3.1, Llama 3.1, GPT OSS, Phi 4, MiniMax M3, Nanbeige, TinyLlama↗
Largest context window
262,144↗
OpenAI-compatible API
Yes↗
Legal entity
Featherless, a Delaware limited liability company↗
Headquarters
United States↗
Founded
2023↗
Support channels
Email, Community, Help-centre↗
Official website
featherless.ai

GPU cloud

GPU models
H100, MI325X, B200, B300↗

Facts checked September 26, 2026, 5 sources

Apps and profiles

Elsewhere
XLinkedIngithub.comhuggingface.coyoutube.com

Pricing

  • Featherless AI Qwen3 Next 80B A3B (USD)

    input-tokens: $0.13 / million

  • Featherless AI Qwen3 Next 80B A3B (USD)

    output-tokens: $1.15 / million

  • Featherless AI Qwen3 Coder 480B (USD)

    input-tokens: $0.38 / million

  • Featherless AI Qwen3 Coder 480B (USD)

    output-tokens: $1.55 / million

  • Featherless AI Qwen3 32B (USD)

    input-tokens: $0.10 / million

  • Featherless AI Qwen3 32B (USD)

    output-tokens: $0.49 / million

  • Featherless AI Qwen3 30B A3B (USD)

    input-tokens: $0.13 / million

  • Featherless AI Qwen3 30B A3B (USD)

    output-tokens: $0.45 / million

  • Featherless AI Qwen3 235B A22B (USD)

    input-tokens: $0.45 / million

  • Featherless AI Qwen3 235B A22B (USD)

    output-tokens: $1.82 / million

  • Featherless AI Qwen 2.5 Coder 32B (USD)

    input-tokens: $0.68 / million

  • Featherless AI Qwen 2.5 Coder 32B (USD)

    output-tokens: $1.20 / million

  • Featherless AI Qwen 2.5 7B (USD)

    input-tokens: $0.17 / million

  • Featherless AI Qwen 2.5 7B (USD)

    output-tokens: $0.20 / million

  • Featherless AI Qwen 2.5 72B (USD)

    input-tokens: $0.37 / million

  • Featherless AI Qwen 2.5 72B (USD)

    output-tokens: $0.40 / million

  • Featherless AI Qwen 2.5 32B (USD)

    input-tokens: $0.68 / million

  • Featherless AI Qwen 2.5 32B (USD)

    output-tokens: $1.20 / million

  • Featherless AI Phi-4 (USD)

    input-tokens: $0.07 / million

  • Featherless AI Phi-4 (USD)

    output-tokens: $0.14 / million

  • Featherless AI Mistral NeMo (USD)

    input-tokens: $0.22 / million

  • Featherless AI Mistral NeMo (USD)

    output-tokens: $0.25 / million

  • Featherless AI MiniMax M2 (USD)

    input-tokens: $0.28 / million

  • Featherless AI MiniMax M2 (USD)

    output-tokens: $1.11 / million

  • Featherless AI Llama 3.3 70B (USD)

    input-tokens: $0.65 / million

  • Featherless AI Llama 3.3 70B (USD)

    output-tokens: $0.75 / million

  • Featherless AI Llama 3.1 8B (USD)

    input-tokens: $0.20 / million

  • Featherless AI Llama 3.1 8B (USD)

    output-tokens: $0.32 / million

  • Featherless AI Kimi K2 (USD)

    input-tokens: $0.60 / million

  • Featherless AI Kimi K2 (USD)

    output-tokens: $2.50 / million

  • Featherless AI gpt-oss-20b (USD)

    input-tokens: $0.04 / million

  • Featherless AI gpt-oss-20b (USD)

    output-tokens: $0.15 / million

  • Featherless AI gpt-oss-120b (USD)

    input-tokens: $0.15 / million

  • Featherless AI gpt-oss-120b (USD)

    output-tokens: $0.60 / million

  • Featherless AI GLM-4.6 (USD)

    input-tokens: $0.55 / million

  • Featherless AI GLM-4.6 (USD)

    output-tokens: $2.20 / million

  • Featherless AI Gemma 3 27B (USD)

    input-tokens: $0.10 / million

  • Featherless AI Gemma 3 27B (USD)

    output-tokens: $0.30 / million

  • Featherless AI Gemma 2 9B (USD)

    input-tokens: $0.43 / million

  • Featherless AI Gemma 2 9B (USD)

    output-tokens: $1.12 / million

  • Featherless AI Gemma 2 27B (USD)

    input-tokens: $0.65 / million

  • Featherless AI Gemma 2 27B (USD)

    output-tokens: $0.65 / million

  • Featherless AI DeepSeek V3.1 (USD)

    input-tokens: $0.29 / million

  • Featherless AI DeepSeek V3.1 (USD)

    output-tokens: $1.00 / million

  • Featherless AI DeepSeek V3 (USD)

    input-tokens: $0.39 / million

  • Featherless AI DeepSeek V3 (USD)

    output-tokens: $1.62 / million

  • Featherless AI DeepSeek R1 Distill (USD)

    input-tokens: $0.04 / million

  • Featherless AI DeepSeek R1 Distill (USD)

    output-tokens: $0.08 / million

  • Featherless AI DeepSeek R1 (USD)

    input-tokens: $0.39 / million

  • Featherless AI DeepSeek R1 (USD)

    output-tokens: $1.62 / million

Pros and cons

Pros

  • The API is OpenAI-compatible, so most existing clients and SDKs keep their current code.
  • The fleet includes H100, MI325X and B200, giving you a spread of model sizes on the same endpoint.
  • The operator is Featherless, a Delaware limited liability company, and the company was founded in 2023.

Cons

  • The largest context window is 262,144, short of the field median of 1,000,000.
  • Support runs on email, community and help centre only, so a production issue waits on email.
  • Output price of $0.02 to $15 goes above the €5.26 reached anywhere else in the field.
Featherless AI

Ready to try Featherless AI?

Scored 9.2 / 10 by DeployCue.

See alternatives

Who it suits

  • Developers on an OpenAI-compatible client who want to change models by editing a model name.
  • Small teams building on open weights such as DeepSeek, Qwen, Mistral and Gemma for chat, coding and extraction.
  • Buyers who want a spread of H100, MI325X and B200 behind the same provider and the same bill.

Who should look elsewhere

  • Not for teams that need a much longer catalogue: Azure AI Foundry serves gpt 6 astra, gpt 6 luna and gpt 6 sol.
  • Not for buyers who want a flat input rate across models: Salad lists $0.06 where this provider runs $0.01 to $3.

Plans, egress and billing

Pricing is metered per request rather than sold as a named plan. Input price runs $0.01 to $3 and output price runs $0.02 to $15, so the same endpoint costs very little for a small open weight model and far more for a large one. Credits are drawn down as you send requests, so the same key covers the whole catalogue.

How Featherless AI compares

Frequently asked questions

How much does it cost to run a model?+

Featherless AI bills per request rather than on a subscription. Input price runs $0.01 to $3 and output price runs $0.02 to $15 across the catalogue, so a small open weight model costs a fraction of a large model on the same endpoint. Credits are drawn down as you send requests.

Which models can I call?+

The catalogue covers DeepSeek 4, DeepSeek 3.2 and GLM 5.3, from small models such as TinyLlama and Nanbeige up to the DeepSeek, Qwen, Kimi, GLM and Mistral families. All of them sit behind the same OpenAI-compatible API, so a prototype can move between them by changing the model name.

How large a context window do I get?+

The largest context window listed is 262,144, short of the field median of 1,000,000. Size prompts to that ceiling instead of assuming the biggest number in the catalogue, and check the model list if a job needs far more.

Do I have to rewrite my code to switch models?+

No rewrite is needed, because the API is OpenAI-compatible. Point an existing client at a different base URL, change the model name, and the same request format covers the whole list. That keeps migration cheap for prototypes that later have to move to another provider.

How do I get help when something breaks?+

Support runs on email, community and help centre, so a live chat or a phone call is not part of the mix. For a production incident you rely on email response times, plus searching the help centre and the community yourself. Keep a fallback model name written into your code in case a model is unavailable.

Who runs Featherless AI and where is it based?+

The operator is Featherless, a Delaware limited liability company, its headquarters country is US, and the company was founded in 2023. A company this young is a judgement call for production workloads, so read the terms of service and decide whether you need your own fallback before you depend on the endpoint.

How does the price compare with the alternatives?+

On entry price it ranks 6 of 71 for input and 5 of 70 for output, so the cheap end is where the value sits. Salad lists $0.06 for input, a similar low rate for a flatter catalogue.

Sources

  • featherless.ai/about checked September 26, 2026

    Backs: Founded, Headquarters

    “2023 Born from open-source research ... and found the company to bring open models into production”

  • featherless.ai/docs/request-pricing-and-credits checked September 26, 2026

    Backs: Input price from, Output price from

    “Prices are listed per 1M tokens ... bge-large-en $0.01 / 1M tok ... kimi3-2780b $3 / 1M tok”

  • featherless.ai/legal/terms-of-service checked September 26, 2026

    Backs: Legal entity

    “between you ("User" or "Developer") and Featherless ("Company", "We", "Us", or "Our"), a Delaware limited liability company”

  • featherless.ai/models checked September 26, 2026

    Backs: Models offered

    “Test any of 40,000+ models effortlessly”

  • featherless.ai/pricing checked September 26, 2026

    Backs: Largest context window, GPU models, OpenAI-compatible API, Support channels

    “Context size up to 256K”

Compiled from 5 source pages, checked September 26, 2026. We did not open an account.

Featherless AI

Featherless AI

Many open weight models behind an OpenAI-compatible endpoint on a tight budget

9.2/10See alternatives