gpt-oss-safeguard-20b vs Phi 4 Mini

Prices checked

gpt-oss-safeguard-20b takes 2 of 5 rounds against Phi 4 Mini. It is also the cheaper of the two to run, at $0.10 per million tokens against $0.13.

OpenAIWins 2 rounds

gpt-oss-safeguard-20b

131k context · tool calling · open weights

$0.10/M tokens

Cheapest price, at Vercel AI Gateway · 5.4¢ per 1,000 pages

Microsoft

Phi 4 Mini

128k context · open weights

$0.13/M tokens

Azure AI Foundry's own price · 6.3¢ per 1,000 pages

Maker's price in (lower wins)

$0.07/M

$0.07/M

Maker's price out (lower wins)

$0.30/M

$0.30/M

Cheapest price (lower wins)

$0.10/M at Vercel AI Gateway

none below the maker

Maker's price, blended (lower wins)

$0.13/M

$0.13/M

Cost per 1,000 pages (lower wins)

5.4¢

6.3¢

Context

131k tokens

128k tokens

Open weights

yes

yes

Tool calling

yes

no

Structured output

yes

no

Reads images

no

no

Reasoning

yes

no

Released

29 Oct 2025

26 Feb 2025

Providers selling it

13

4

Price we measure against

OpenRouter

Azure AI Foundry

Trains on your prompts (no is better)

not published

no provider to check

Providers selling both · 1

  • NanoGPT$0.13 / $0.30

gpt-oss-safeguard-20b, then Phi 4 Mini, per million tokens.

Go further

Add a third or fourth model, or change the weighting, in the tool.

Frequently asked

Which is cheaper, gpt-oss-safeguard-20b or Phi 4 Mini?
gpt-oss-safeguard-20b, at $0.10 per million tokens against $0.13 for Phi 4 Mini, blended 3 in : 1 out, read on 21 Oct 2026.
Where is each one cheapest?
gpt-oss-safeguard-20b: $0.10 at Vercel AI Gateway. Phi 4 Mini: nobody sells it below Azure AI Foundry's own price with a stated reason, so $0.13 is the number to beat.
Can I buy both from one provider?
Yes. One provider sells both: NanoGPT. The cheapest of them for the two together is NanoGPT.
Which holds more context?
gpt-oss-safeguard-20b, at 131k tokens against 128k.

Prices exclude tax, read 21 Oct 2026 from each provider's own catalog. They are per M tokens, blended 3 in : 1 out, and a page is 600 tokens in and 60 out. Bars are scaled to the larger value in each row.

Related comparisons

More