# DeepInfra

36 Hacker News commenters recommended DeepInfra and 11 criticized it, November 2025 to November 2026. It was recommended in 11 of 13 months. It meets the Hacker Approved badge rule.

**Verdict:** Recommended for cheap, fast access to many open-source models through an OpenAI-compatible API. Critics say it quantizes some models and sometimes runs them worse.

## Why people recommend it

- Cheap prices for open models (9 comments): "offer a wide range of models like gpt oss that are very capable and far cheaper than the models from big 3." ([source](https://news.ycombinator.com/item?id=48435889)) "I use GLM 4.7 with DeepInfra.com and it's extremely reasonable, though maybe a bit on the slower side." ([source](https://news.ycombinator.com/item?id=46686303)) "run the same open models for far cheaper and faster than you could ever build out at home" ([source](https://news.ycombinator.com/item?id=46355283))
- Wide range of open models (5 comments): "offer a wide range of models like gpt oss that are very capable and far cheaper than the models from big 3." ([source](https://news.ycombinator.com/item?id=48435889)) "try Fireworks AI or DeepInfra who do support more recent models" ([source](https://news.ycombinator.com/item?id=49157760)) "because they host open-source models that are good." ([source](https://news.ycombinator.com/item?id=47618474))
- Fast throughput on most models (3 comments): "I'm very happy with Deepinfra. Less model coverage, but good prices and quite fast." ([source](https://news.ycombinator.com/item?id=49697167)) "the prices are decent for US-based deployments and there's acceptable throughput" ([source](https://news.ycombinator.com/item?id=48814137)) "I pretty reliably get 200 tokens per second on deep infra." ([source](https://news.ycombinator.com/item?id=50238382))
- US-based with good terms (4 comments): "the prices are decent for US-based deployments and there's acceptable throughput" ([source](https://news.ycombinator.com/item?id=48814137)) "Their whole catalogue is openai compat. US based." ([source](https://news.ycombinator.com/item?id=49129728)) "DeepInfra has excellent terms of service too!" ([source](https://news.ycombinator.com/item?id=49483944))

## What people warn about

- Quantized models, possibly lower quality (4 comments): "Some of them more heavily quantize their models (deepinfra) so take that into your workload considerations." ([source](https://news.ycombinator.com/item?id=50130493)) "it's at FP4 quant. How much that effects the quality of inference for GLM, I can't say." ([source](https://news.ycombinator.com/item?id=48990725)) "has a bad history of quantizing models" ([source](https://news.ycombinator.com/item?id=50019783))
- Does not preserve model thinking (2 comments): "Deepinfra's implementation of it is not correct. Thinking is not preserved, and they're not responding to my submitted issue about it." ([source](https://news.ycombinator.com/item?id=47835520)) "Deepinfra for example is not preserving thinking correctly for GLM5.1" ([source](https://news.ycombinator.com/item?id=47838491))
- Slower than some providers (2 comments): "I use GLM 4.7 with DeepInfra.com and it's extremely reasonable, though maybe a bit on the slower side." ([source](https://news.ycombinator.com/item?id=46686303)) "deepinfra.com is cheaper for that but its noticeably slower" ([source](https://news.ycombinator.com/item?id=50215603))
- Weak cache hit rate and pricing (2 comments): "their cache hit rate is 67%. In comparison the provider with the highest hit rate is at 95%." ([source](https://news.ycombinator.com/item?id=49482025)) "DeepInfra's Cache Read is 6.5 times more expensive than DeepSeek." ([source](https://news.ycombinator.com/item?id=49154023))

## Alternatives hackers recommend

- [Together AI](https://hackerapproved.com/p/together-ai.md): 20 recommend
- [fal.ai](https://hackerapproved.com/p/fal-ai.md): 19 recommend
- [Neuralwatt](https://hackerapproved.com/p/neuralwatt.md): 18 recommend
- [Baseten](https://hackerapproved.com/p/baseten.md): 15 recommend
- [Replicate](https://hackerapproved.com/p/replicate.md): 13 recommend

## Where to get it

- [Visit the official site](https://deepinfra.com/)

## Cite as

Hacker Approved, "DeepInfra", https://hackerapproved.com/p/deepinfra/ (counts from 1,795,640 Hacker News comments, November 2025 to November 2026).